WEBVTT 00:00:00.000 --> 00:00:01.740 You see these AI apps and stuff like that, that 00:00:01.740 --> 00:00:03.700 people use. And I think the programs that they 00:00:03.700 --> 00:00:05.459 build on some of the better ones are like perfectly 00:00:05.459 --> 00:00:07.740 fine. You know what I mean? But you know, you're 00:00:07.740 --> 00:00:09.660 never, you're not going to get like, it's not 00:00:09.660 --> 00:00:11.240 going to help you on a competition day. It's 00:00:11.240 --> 00:00:12.980 not going to critique your technique. It's not 00:00:12.980 --> 00:00:14.599 going to help you with your mindset and stuff 00:00:14.599 --> 00:00:16.179 like that as well when it comes to approaching 00:00:16.179 --> 00:00:18.600 competition. And I think that that is something 00:00:18.600 --> 00:00:20.839 very similar that I would highlight with respect 00:00:20.839 --> 00:00:22.820 to our model in the sense that this will never 00:00:22.820 --> 00:00:25.199 plan a competition strategy for you. It will 00:00:25.199 --> 00:00:27.500 just maybe give you some information that is 00:00:27.500 --> 00:00:30.179 useful in your coach's book. it alongside everything 00:00:30.179 --> 00:00:32.859 else to help you make the application of the 00:00:32.859 --> 00:00:34.960 art of coaching a little more informed and perhaps 00:00:34.960 --> 00:00:37.460 a little more systematic if possible you know 00:00:37.460 --> 00:00:39.939 that that's kind of the intent behind it at least 00:00:41.740 --> 00:00:44.000 Hi Ian, it's my pleasure to have you on Evidence 00:00:44.000 --> 00:00:46.240 Strong Show. If you could briefly introduce yourself. 00:00:46.500 --> 00:00:48.359 Hi Alex, thank you very much for having me on. 00:00:48.420 --> 00:00:51.020 It's my pleasure. My name is Ian Darrah. I guess 00:00:51.020 --> 00:00:53.299 I have a bit of a mixed background. So I have 00:00:53.299 --> 00:00:56.600 a PhD in exercise physiology. My PhD was mostly 00:00:56.600 --> 00:00:59.159 focused kind of around the domain of trying to 00:00:59.159 --> 00:01:02.399 identify almost different biomarkers of different 00:01:02.399 --> 00:01:04.500 training status. So resistance trained, endurance 00:01:04.500 --> 00:01:06.840 trained individuals. Before kind of getting into 00:01:06.840 --> 00:01:09.500 sports science, I was an athlete. So I was a 00:01:09.500 --> 00:01:12.719 reasonably decent. national level Olympic weightlifter 00:01:12.719 --> 00:01:15.000 in Ireland. I did a couple of international competitions 00:01:15.000 --> 00:01:17.260 as a junior as well. A couple of junior national 00:01:17.260 --> 00:01:19.140 medals, a couple of senior national medals. So 00:01:19.140 --> 00:01:21.760 like an OK strength athlete, which got my interest 00:01:21.760 --> 00:01:24.099 in sports science. When I was an undergrad, up 00:01:24.099 --> 00:01:26.219 until still kind of now, I did a lot of coaching 00:01:26.219 --> 00:01:28.700 as well. So a lot of Olympic weightlifting coaching 00:01:28.700 --> 00:01:31.060 for probably about 10 years, a small bit of strength 00:01:31.060 --> 00:01:33.420 and conditioning stuff as well. And I now currently 00:01:33.420 --> 00:01:36.819 work for a large hospital network here in northeast 00:01:36.819 --> 00:01:39.719 Pennsylvania. I work as a researcher in a sports 00:01:39.719 --> 00:01:41.920 medicine department. kind of essentially working 00:01:41.920 --> 00:01:44.819 with large scale electronic health record data 00:01:44.819 --> 00:01:46.879 to do looking at the influence at the moment 00:01:46.879 --> 00:01:49.640 now, mostly of malnutrition on surgical outcomes. 00:01:49.900 --> 00:01:51.939 So again, kind of integrating some of that physiology 00:01:51.939 --> 00:01:54.439 stuff and nutrition stuff a little bit with some 00:01:54.439 --> 00:01:57.459 data science too. And the paper is about powerlifting. 00:01:57.540 --> 00:02:00.620 So how did it come to be? Open Powerlifting is 00:02:00.620 --> 00:02:03.599 an extremely well -maintained powerlifting database. 00:02:03.780 --> 00:02:05.379 I'm sure many of your lifters, particularly anyone 00:02:05.379 --> 00:02:07.379 who competes in powerlifting are familiar with 00:02:07.379 --> 00:02:10.310 it. But whoever runs that website, I don't actually 00:02:10.310 --> 00:02:12.509 know the individuals by name or anything like 00:02:12.509 --> 00:02:15.090 that. They do an extremely good job of kind of 00:02:15.090 --> 00:02:17.090 curating the database and maintaining it. If 00:02:17.090 --> 00:02:19.270 I recall correctly, I think it's updated potentially 00:02:19.270 --> 00:02:21.469 daily. If it's not updated daily, it's updated 00:02:21.469 --> 00:02:24.189 weekly. And they make that data free and open 00:02:24.189 --> 00:02:26.830 source and available for download as a CSV file. 00:02:26.949 --> 00:02:29.030 So that was how I kind of came up with the idea 00:02:29.030 --> 00:02:31.870 of working with this database. And I kind of 00:02:31.870 --> 00:02:34.610 iterated through a few different ideas, trying 00:02:34.610 --> 00:02:36.110 to come up with something, partially looking 00:02:36.110 --> 00:02:38.229 at the literature, partially just kind of playing 00:02:38.229 --> 00:02:39.750 around with different ideas. for a couple of 00:02:39.750 --> 00:02:41.469 months. And then believe it or not, I actually 00:02:41.469 --> 00:02:43.830 came across an Instagram post. It was just a 00:02:43.830 --> 00:02:45.310 spot. I can't remember who posted it, but it 00:02:45.310 --> 00:02:47.090 was just a small little infographic that was 00:02:47.090 --> 00:02:49.930 talking about how much you jump between lifts 00:02:49.930 --> 00:02:52.229 and basically strategizing, you know, when you're 00:02:52.229 --> 00:02:54.430 in a powerlifting meet, how much should you jump 00:02:54.430 --> 00:02:56.289 between first attempt, second attempt, so on 00:02:56.289 --> 00:02:58.409 and so forth. And I was like, OK, this is actually 00:02:58.409 --> 00:03:00.189 something we could potentially look at here. 00:03:00.310 --> 00:03:03.389 So the idea that I kind of crudely came up with 00:03:03.389 --> 00:03:06.449 was I have this this database. I kind of diluted 00:03:06.449 --> 00:03:08.979 it down into some specific. specific contexts. 00:03:09.020 --> 00:03:11.300 And I was like, let's see if we can potentially 00:03:11.300 --> 00:03:13.919 quantify or at least explore with some degree 00:03:13.919 --> 00:03:17.500 of detail the relationship between how much you 00:03:17.500 --> 00:03:19.659 decide to jump between two subsequent attempts. 00:03:19.800 --> 00:03:22.020 So if my opening attempt is 100 kilograms and 00:03:22.020 --> 00:03:24.680 I take my second attempt at 110 kilograms, I've 00:03:24.680 --> 00:03:27.280 jumped 10 kilograms. Can we look at how that 00:03:27.280 --> 00:03:30.300 influences the probability that I will be successful 00:03:30.300 --> 00:03:32.699 on the attempt that I'm jumping to? So if I make 00:03:32.699 --> 00:03:35.060 that 10 kilo jump, what is the probability that 00:03:35.060 --> 00:03:37.860 that attempt will be successful? This is perfect 00:03:37.860 --> 00:03:40.740 because they're predicting or they're trying 00:03:40.740 --> 00:03:44.159 to come up with reasonable attempts and jumps 00:03:44.159 --> 00:03:46.719 for the competition. This is part of the art 00:03:46.719 --> 00:03:49.319 of coaching. So the coach develops relationship 00:03:49.319 --> 00:03:53.439 with the athletes so that in part they can predict 00:03:53.439 --> 00:03:57.319 when the athlete is more likely to make the next 00:03:57.319 --> 00:04:00.139 attempt and what the jumps should be. And here 00:04:00.139 --> 00:04:03.219 you are taking a big data set from many, many 00:04:03.219 --> 00:04:05.879 athletes, thousands of athletes and trying to 00:04:05.879 --> 00:04:09.580 see. whether we can actually statistically calculate 00:04:09.580 --> 00:04:14.280 what the jumps are. And if we do a specific jump, 00:04:14.400 --> 00:04:17.180 what is the probability of hitting the attempt 00:04:17.180 --> 00:04:21.139 for females and for males? Yeah, and that's probably 00:04:21.139 --> 00:04:23.480 something to touch on in that one of the things 00:04:23.480 --> 00:04:25.259 that's really important to mention is that for 00:04:25.259 --> 00:04:28.279 predictive models to be useful, they often need 00:04:28.279 --> 00:04:30.620 a really large amount of data. One of the things 00:04:30.620 --> 00:04:32.199 I've commented on as well is the fact that we 00:04:32.199 --> 00:04:35.060 filtered it down to IPF, International Powerlifting 00:04:35.060 --> 00:04:37.250 Federation. competitions for a couple of reasons 00:04:37.250 --> 00:04:40.170 but one of them is that they have very strict 00:04:40.170 --> 00:04:43.149 technical rulings so you have i think in the 00:04:43.149 --> 00:04:46.910 study we had around 96 or 93 000 lifters total 00:04:46.910 --> 00:04:49.170 you know which is an insane number of competitors 00:04:49.170 --> 00:04:53.189 but because of the rule set we can also say within 00:04:53.189 --> 00:04:56.589 a reason like within reason that they the lifts 00:04:56.589 --> 00:04:59.629 were judged under somewhat consistent standards 00:04:59.629 --> 00:05:02.269 you know could you explain what building a model 00:05:02.269 --> 00:05:05.290 is i can talk in detail about particular type 00:05:05.290 --> 00:05:07.389 of model we did maybe later down the line but 00:05:07.389 --> 00:05:09.410 what you essentially see as a model and you'll 00:05:09.410 --> 00:05:11.430 see them in research studies they are in essence 00:05:11.430 --> 00:05:14.069 something kind of like a generalizable equation 00:05:14.069 --> 00:05:17.089 so what i mean by that is the sense that if we 00:05:17.089 --> 00:05:18.870 have built this model today and we've reported 00:05:18.870 --> 00:05:20.970 the equation of it there are a bunch of factors 00:05:20.970 --> 00:05:23.310 in that equation such as the body weight of the 00:05:23.310 --> 00:05:25.790 lifter the age of the lifter the sex of the lifter 00:05:25.790 --> 00:05:28.029 and so on as well as some constant values and 00:05:28.029 --> 00:05:31.069 the idea there is that it's kind of like like 00:05:31.069 --> 00:05:33.230 a calculator essentially in that if you took 00:05:33.230 --> 00:05:35.519 that equation and you You had values for yourself. 00:05:35.839 --> 00:05:38.199 You know, I know my age. I know my sex. I know 00:05:38.199 --> 00:05:39.899 my body weight. I know my jump attempt size. 00:05:40.100 --> 00:05:41.620 You can plug them in and you'll get a result, 00:05:41.740 --> 00:05:44.120 you know, just by using the constants. So a model 00:05:44.120 --> 00:05:46.920 is essentially like a pre -built calculation. 00:05:47.319 --> 00:05:50.920 You know, it has inputs and you again, like it 00:05:50.920 --> 00:05:53.300 has inputs and it has an output. And obviously 00:05:53.300 --> 00:05:55.379 the output is generally what you're kind of trying 00:05:55.379 --> 00:05:57.360 to, you know, the desirable thing you're trying 00:05:57.360 --> 00:05:59.300 to predict if it's a predictive model. And then 00:05:59.300 --> 00:06:02.399 models also have essentially quality. You know, 00:06:02.439 --> 00:06:04.459 you have good model. Good models and bad models. 00:06:04.560 --> 00:06:07.220 There's a very famous statistics quote. I can't 00:06:07.220 --> 00:06:08.819 remember who it's attributed to, but it's the 00:06:08.819 --> 00:06:11.540 idea that all models are terrible, but some models 00:06:11.540 --> 00:06:13.800 are useful. So something to kind of note is that 00:06:13.800 --> 00:06:16.639 there will always be a detraction from reality 00:06:16.639 --> 00:06:18.959 in any statistical model that you build. So even 00:06:18.959 --> 00:06:20.819 if you see something that is, and there are a 00:06:20.819 --> 00:06:22.660 number of parameters, essentially, you'll build 00:06:22.660 --> 00:06:24.639 a statistical model, you'll assess its quality 00:06:24.639 --> 00:06:26.860 based on a number of parameters. So you can tell, 00:06:26.939 --> 00:06:28.720 is this model good at predicting the outcome 00:06:28.720 --> 00:06:30.660 or bad at predicting the outcome in essence? 00:06:30.819 --> 00:06:32.699 But even models that are very good at predicting 00:06:32.699 --> 00:06:34.540 the outcome. predicting outcomes are not perfect 00:06:34.540 --> 00:06:36.579 you'll never have a model that particularly in 00:06:36.579 --> 00:06:38.740 this circumstance let's say i wouldn't call it 00:06:38.740 --> 00:06:40.879 a criticism but one of the skepticisms we've 00:06:40.879 --> 00:06:42.500 received from kind of friends of mine who are 00:06:42.500 --> 00:06:44.540 practitioners and stuff like that is the idea 00:06:44.540 --> 00:06:46.939 that oh so you think that like this will make 00:06:46.939 --> 00:06:49.100 a decision for coaches and the idea that is like 00:06:49.100 --> 00:06:51.160 absolutely not because there is still a black 00:06:51.160 --> 00:06:53.680 box component here this is more about understanding 00:06:53.680 --> 00:06:56.540 the relationships and giving some maybe quantifiable 00:06:56.540 --> 00:06:59.060 benchmarks that you can come in with some baseline 00:06:59.060 --> 00:07:01.620 information and then add that to what you intuitively 00:07:01.930 --> 00:07:03.550 know about your athlete through working with 00:07:03.550 --> 00:07:05.569 them or just, again, through your coach's intuition. 00:07:05.910 --> 00:07:08.149 It's about informing decisions instead of making 00:07:08.149 --> 00:07:10.230 them for you. So I've kind of gone on a bit of 00:07:10.230 --> 00:07:12.589 a tangent there, but maybe to bring it back in 00:07:12.589 --> 00:07:15.009 together, a model is essentially, it's a machine. 00:07:15.170 --> 00:07:17.850 It has a number of inputs and it's built and 00:07:17.850 --> 00:07:20.029 you'll build a model sequentially. So you'll 00:07:20.029 --> 00:07:22.110 start with one input. So for example, when we 00:07:22.110 --> 00:07:24.750 developed the model for this study, we had the 00:07:24.750 --> 00:07:27.500 output decided, which was... whether a lift would 00:07:27.500 --> 00:07:29.459 be successful or not and then we had our first 00:07:29.459 --> 00:07:31.579 input which was the jump size between attempts 00:07:31.579 --> 00:07:34.120 so we just started with two things the jump size 00:07:34.120 --> 00:07:36.459 whether it's 10 kilograms 20 kilograms whatever 00:07:36.459 --> 00:07:38.779 and the probability of a successful attempt and 00:07:38.779 --> 00:07:40.259 we looked at that and we go okay how well is 00:07:40.259 --> 00:07:42.759 how good is this alone at predicting our outcome 00:07:42.759 --> 00:07:44.779 and then we added something in well what about 00:07:44.779 --> 00:07:47.360 whether it's different between squat bench and 00:07:47.360 --> 00:07:49.279 deadlifts because the amount you will jump is 00:07:49.279 --> 00:07:50.980 different between whether the lift is a squat 00:07:50.980 --> 00:07:53.120 bench or deadlift so then we added in the lift 00:07:53.120 --> 00:07:55.879 type and that was a factor then we added in What 00:07:55.879 --> 00:07:57.899 we've termed the attempt transition, which is 00:07:57.899 --> 00:07:59.759 not a great term, but that's essentially whether 00:07:59.759 --> 00:08:01.680 you're jumping between the first and second attempt 00:08:01.680 --> 00:08:04.399 or the second and third attempt. So how far you 00:08:04.399 --> 00:08:06.680 are within each given lift, you add that in and 00:08:06.680 --> 00:08:09.500 you can see if nuance kind of is extracted further 00:08:09.500 --> 00:08:11.259 on down the lines. That's essentially how the 00:08:11.259 --> 00:08:13.920 model was built by sequentially adding in different 00:08:13.920 --> 00:08:16.060 factors that we had data on that we thought were 00:08:16.060 --> 00:08:18.040 relevant, evaluating the model's parameters, 00:08:18.240 --> 00:08:20.420 but then also trying to balance it off how easy 00:08:20.420 --> 00:08:22.399 it would be to interpret the model. So there 00:08:22.399 --> 00:08:24.319 are other factors we could have added to our 00:08:24.319 --> 00:08:27.089 model. make it a better predictor, but it becomes 00:08:27.089 --> 00:08:29.290 too confusing to understand how it's producing 00:08:29.290 --> 00:08:31.269 results at that point, which means it's more 00:08:31.269 --> 00:08:34.509 difficult to use those results in practice. So 00:08:34.509 --> 00:08:36.509 again, that's kind of that trade -off between 00:08:36.509 --> 00:08:39.409 complexity and interpretability that I was talking 00:08:39.409 --> 00:08:41.169 about at the beginning. My understanding for 00:08:41.169 --> 00:08:44.669 now is you found this data set, you decided to 00:08:44.669 --> 00:08:47.090 look at the jumps and the probability of success, 00:08:47.350 --> 00:08:51.269 and then you tried to include different variables 00:08:51.269 --> 00:08:53.690 one by one, adding them to the model and seeing 00:08:53.690 --> 00:08:56.250 what... you can, that the results will be different 00:08:56.250 --> 00:08:59.149 and whether the model is a good fit. How do you 00:08:59.149 --> 00:09:02.169 call it? Yeah, essentially, like, does it become 00:09:02.169 --> 00:09:04.789 better at predicting whether a lift will be successful 00:09:04.789 --> 00:09:07.470 or not? And I mean, I think contextualizing that 00:09:07.470 --> 00:09:09.629 probably makes it a little bit easier to understand 00:09:09.629 --> 00:09:12.429 in the sense of, if we think about just considering 00:09:12.429 --> 00:09:15.029 how much you jump alone, you know, that can be 00:09:15.029 --> 00:09:17.710 useful. But the amount you jump, it doesn't matter 00:09:17.710 --> 00:09:20.830 if you're a six foot eight, 200 kilogram male 00:09:20.830 --> 00:09:24.259 or a four foot nine, 48 kilo. female the amount 00:09:24.259 --> 00:09:26.639 you jump will always be bigger on your squat 00:09:26.639 --> 00:09:28.360 and your deadlift than it will be on your bench 00:09:28.360 --> 00:09:30.759 press you know like a you know for some people 00:09:30.759 --> 00:09:33.460 a 50 kilogram jump on the squat may be achievable 00:09:33.460 --> 00:09:35.700 but most people will not make a 50 kilogram jump 00:09:35.700 --> 00:09:37.580 on the bench press so if you go oh well this 00:09:37.580 --> 00:09:39.899 model only considers how much you jump but the 00:09:39.899 --> 00:09:42.320 amount i jump is very very different between 00:09:42.320 --> 00:09:44.879 my bench press let's say than my deadlift you 00:09:44.879 --> 00:09:47.759 know or the amount i jump is very very different 00:09:47.759 --> 00:09:50.720 between my last opening or my last warm -up weight 00:09:50.720 --> 00:09:53.759 and my opening weight you know Or my third attempt 00:09:53.759 --> 00:09:56.320 is so on and so forth. So it's just like the 00:09:56.320 --> 00:09:59.960 best way to think about it is like because like 00:09:59.960 --> 00:10:01.539 the way you kind of think about it when you're 00:10:01.539 --> 00:10:03.419 building the model is, oh, but what about this 00:10:03.419 --> 00:10:05.879 situation? And do I have a variable that I can 00:10:05.879 --> 00:10:07.559 contextualize that to? You know what I mean? 00:10:07.600 --> 00:10:10.139 But what about older lifters versus younger lifters? 00:10:10.259 --> 00:10:12.419 Does the effect stay for older lifters? OK, I 00:10:12.419 --> 00:10:14.480 should include age as a variable in the equation 00:10:14.480 --> 00:10:16.720 to see if that changes the prediction. What about, 00:10:16.799 --> 00:10:19.340 you know, males versus females? What about squats 00:10:19.340 --> 00:10:21.399 versus bench press and so on and so forth? So 00:10:21.399 --> 00:10:23.820 that. The main predictor is this jump size, but 00:10:23.820 --> 00:10:25.720 then we have all these other variables. And the 00:10:25.720 --> 00:10:28.100 statistical term is a covariate, but it's essentially 00:10:28.100 --> 00:10:31.320 like caveat, a contextualizing factor, you know. 00:10:31.320 --> 00:10:33.159 But obviously, the more you contextualize, the 00:10:33.159 --> 00:10:35.639 more difficult it can become to just generalize. 00:10:35.659 --> 00:10:37.379 You know what I mean? The more nuanced you make 00:10:37.379 --> 00:10:39.279 your model, the more difficult it is to say how 00:10:39.279 --> 00:10:41.220 it applies to everybody, you know, to some degree. 00:10:41.419 --> 00:10:43.340 And so there is, as I've kind of mentioned, a 00:10:43.340 --> 00:10:46.100 balancing act between, you know, making sure 00:10:46.100 --> 00:10:48.740 that you explain enough that the information 00:10:48.740 --> 00:10:50.720 is useful, but you don't explain so much that. 00:10:51.049 --> 00:10:53.610 You can't say, like, you can't give a statement 00:10:53.610 --> 00:10:56.789 that most people can then take information from 00:10:56.789 --> 00:10:58.490 to use in their, in their training or their, 00:10:58.549 --> 00:11:00.769 their competing practices. Essentially what we 00:11:00.769 --> 00:11:05.149 used is a, what's known as a non -linear logistic 00:11:05.149 --> 00:11:08.009 regression approach. And again, you'll see this 00:11:08.009 --> 00:11:09.909 with statistics all the time and that like you 00:11:09.909 --> 00:11:12.230 get all these fancy terms, but it is actually 00:11:12.230 --> 00:11:14.490 something that can be simplified. So regression 00:11:14.490 --> 00:11:17.009 is a term that you see thrown around a lot. Anytime 00:11:17.009 --> 00:11:18.990 you see the word regression in relation to a 00:11:18.990 --> 00:11:20.549 statistical model, it's simple. simply means 00:11:20.549 --> 00:11:22.190 a model that's looking to predict something. 00:11:22.610 --> 00:11:25.210 Logistic then simply refers to the fact that 00:11:25.210 --> 00:11:27.850 it's using a logic function, which basically 00:11:27.850 --> 00:11:31.269 is designed to look at binary variables. So yes, 00:11:31.389 --> 00:11:34.309 no. So in our case, was the lift successful or 00:11:34.309 --> 00:11:36.590 not? You know, a lift cannot be kind of successful. 00:11:36.769 --> 00:11:38.330 You know, the judges either give you two white 00:11:38.330 --> 00:11:40.330 lights or three white lights or they don't. You 00:11:40.330 --> 00:11:42.909 know, so when we say logistic regression, it's 00:11:42.909 --> 00:11:45.169 a model that's designed to predict a yes, no 00:11:45.169 --> 00:11:47.610 outcome. Good lift, bad lift. That's kind of 00:11:47.610 --> 00:11:50.419 the logistic element of it. Non -linear, how 00:11:50.419 --> 00:11:53.360 regression works in its most simple form is that 00:11:53.360 --> 00:11:56.480 you have your data already collected or provided 00:11:56.480 --> 00:12:00.620 to you or whatever, and you come in with a preconceived 00:12:00.620 --> 00:12:03.399 relationship between your variables. So probably 00:12:03.399 --> 00:12:05.740 the best example to think about is linear regression, 00:12:05.980 --> 00:12:07.799 which even if you've never heard that term before, 00:12:07.919 --> 00:12:10.019 you've probably actually seen it. If you've ever 00:12:10.019 --> 00:12:11.799 seen a scatter plot with a bunch of different 00:12:11.799 --> 00:12:14.279 points and exactly a slanted line that's either 00:12:14.279 --> 00:12:16.840 going upwards or downwards, that's linear regression. 00:12:17.080 --> 00:12:19.679 And how linear regression works essentially is 00:12:19.679 --> 00:12:22.299 I say, I think there is a straight line relationship 00:12:22.299 --> 00:12:24.759 between my two variables that I'm interested 00:12:24.759 --> 00:12:28.320 in. Let's say the amount I jump in a lift and 00:12:28.320 --> 00:12:30.279 the probability of that lift being successful. 00:12:30.419 --> 00:12:32.399 I think the more you jump, the more likely you 00:12:32.399 --> 00:12:34.559 are to be successful or the more you jump, the 00:12:34.559 --> 00:12:37.179 less likely you are to be successful. And you 00:12:37.179 --> 00:12:40.200 draw that line the best you can based on how 00:12:40.200 --> 00:12:42.879 the data is shaped. And the quality of the model 00:12:42.879 --> 00:12:45.500 is essentially assessed based on how well your 00:12:45.500 --> 00:12:48.440 data orbits around that straight line. So if 00:12:48.440 --> 00:12:51.500 I drew a straight line for that example and all 00:12:51.500 --> 00:12:53.720 the data was just very, very close to the line, 00:12:53.779 --> 00:12:55.340 we'd be like, oh, this is a great model because 00:12:55.340 --> 00:12:57.940 this shape that I've drawn, which is explaining 00:12:57.940 --> 00:13:00.639 a relationship, a very linear relationship, more 00:13:00.639 --> 00:13:03.600 is better, less or less is worse. My data conforms 00:13:03.600 --> 00:13:05.580 to that shape. Therefore, it's a good model. 00:13:05.679 --> 00:13:08.240 We used a nonlinear approach. And the reason 00:13:08.240 --> 00:13:10.379 for that is a lot of relationships are nonlinear. 00:13:10.460 --> 00:13:13.460 And all nonlinear means is the line is not straight. 00:13:13.639 --> 00:13:15.820 And there are actually plenty of examples in 00:13:15.820 --> 00:13:18.029 sports and exercise science. of nonlinear relationships 00:13:18.029 --> 00:13:20.070 that a lot of people might be familiar with. 00:13:20.190 --> 00:13:22.250 Anyone who's ever seen the results of a VO2 max 00:13:22.250 --> 00:13:24.750 test before, so essentially an aerobic fitness 00:13:24.750 --> 00:13:26.769 test where you run on a treadmill, you ride a 00:13:26.769 --> 00:13:29.210 bike, usually for 12 to 15 minutes, the test 00:13:29.210 --> 00:13:31.970 gets progressively harder minute on minute and 00:13:31.970 --> 00:13:34.049 you track oxygen consumption. What you'll see 00:13:34.049 --> 00:13:36.330 is oxygen consumption increases in a straight 00:13:36.330 --> 00:13:38.470 line and then it reaches a plateau and it stops. 00:13:38.610 --> 00:13:40.929 So it's almost kind of like a truncated S shape 00:13:40.929 --> 00:13:43.370 or something like that. That is an example of 00:13:43.370 --> 00:13:45.330 a nonlinear relationship. Oxygen consumption 00:13:45.330 --> 00:13:47.370 increases, increases. increases when you reach 00:13:47.370 --> 00:13:50.049 302 max it gets stuck so that that's an example 00:13:50.049 --> 00:13:52.169 of one we see in sports and exercise science 00:13:52.169 --> 00:13:54.350 and we did the same thing here so we said there 00:13:54.350 --> 00:13:57.350 is a we're building a model to predict this outcome 00:13:57.350 --> 00:13:59.730 of good lift or bad lift and we're assuming that 00:13:59.730 --> 00:14:01.909 this relationship is non -linear basically meaning 00:14:01.909 --> 00:14:05.070 that there will be some kind of curve in the 00:14:05.070 --> 00:14:07.230 relationship and we can kind of get in when we 00:14:07.230 --> 00:14:08.990 get into the results it's a better way to kind 00:14:08.990 --> 00:14:11.649 of explain it but like contextualizing it the 00:14:11.649 --> 00:14:13.409 best way you could describe it is increasing 00:14:13.409 --> 00:14:16.399 your jump size to some extent will potentially 00:14:16.399 --> 00:14:19.360 increase your probability of a successful lift 00:14:19.360 --> 00:14:21.559 up into a point. But obviously you can't jump 00:14:21.559 --> 00:14:23.500 forever. You know what I mean? So if you say, 00:14:23.639 --> 00:14:27.440 oh, if I jump two kilos versus five kilos, I 00:14:27.440 --> 00:14:29.220 might have a better chance of making the lift 00:14:29.220 --> 00:14:31.759 with five kilos. But you can't do that into infinity 00:14:31.759 --> 00:14:33.759 where you go, well, then, well, if I jump eight 00:14:33.759 --> 00:14:36.500 kilos, well, if I jump 18 kilos, well, if I jump 00:14:36.500 --> 00:14:38.620 25 kilos, anyone who's ever lifted before is 00:14:38.620 --> 00:14:40.460 going to go, okay, that doesn't make sense. You 00:14:40.460 --> 00:14:42.879 know, there comes a point. I add 200 kilos, you 00:14:42.879 --> 00:14:45.200 know, it eventually levels off. That's essentially 00:14:45.200 --> 00:14:48.220 what the non -linearity aspect of our model is 00:14:48.220 --> 00:14:50.259 kind of meant to describe. The fact that like 00:14:50.259 --> 00:14:54.220 you may see an increase up into a point in benefit, 00:14:54.340 --> 00:14:57.080 but through logic, there has to be a ceiling 00:14:57.080 --> 00:15:00.080 where the relationship begins to kind of change. 00:15:00.220 --> 00:15:04.340 And so in its most simple form, think of the 00:15:04.340 --> 00:15:06.860 relationship between jump size and the probability 00:15:06.860 --> 00:15:09.220 of lift being successful is a squiggly line. 00:15:09.559 --> 00:15:11.340 of some description. It's not a straight line 00:15:11.340 --> 00:15:13.940 relationship. And again, that just conveys a 00:15:13.940 --> 00:15:16.559 little bit of complexity there, less so to do 00:15:16.559 --> 00:15:18.759 with the model and more so to do with the phenomenon 00:15:18.759 --> 00:15:20.960 itself. But I think anyone who's lifted before 00:15:20.960 --> 00:15:24.360 will have some intuitive feeling of that. Anyone 00:15:24.360 --> 00:15:26.580 who's actually loaded a bar up to a one rep max 00:15:26.580 --> 00:15:28.740 or a three rep max or competed in some kind of 00:15:28.740 --> 00:15:31.379 way will understand that feeling to some extent. 00:15:31.580 --> 00:15:33.519 So there is some logic behind it, which I think 00:15:33.519 --> 00:15:35.299 makes it a small bit easier to understand. But 00:15:35.299 --> 00:15:38.820 I do realize that element of the study is core. 00:15:38.860 --> 00:15:40.860 quite complex and difficult to digest for sure. 00:15:41.629 --> 00:15:45.029 No, I think the practicality of the study is 00:15:45.029 --> 00:15:47.990 at some point it makes no sense to make this 00:15:47.990 --> 00:15:51.570 jump. So the jump is too big or it can be also 00:15:51.570 --> 00:15:54.529 too small, but let's focus on too big. So if 00:15:54.529 --> 00:15:57.649 you are coaching through the competition an athlete 00:15:57.649 --> 00:16:01.649 you do not know personally or your relationship 00:16:01.649 --> 00:16:05.450 with this athlete is very short so that you don't 00:16:05.450 --> 00:16:09.169 have experience in jumps, there is probably a 00:16:09.169 --> 00:16:12.129 number for this particular. athlete or because 00:16:12.129 --> 00:16:15.490 they are a male in this weight class and it's 00:16:15.490 --> 00:16:18.549 a back squat there is a cut of number that you 00:16:18.549 --> 00:16:22.370 shouldn't cross in this generic let's say situation 00:16:22.370 --> 00:16:24.990 only because you don't have this experience and 00:16:24.990 --> 00:16:27.269 relationship developed one of the things i will 00:16:27.269 --> 00:16:29.570 say is like i think definitely take anything 00:16:29.570 --> 00:16:31.870 when we talk about specific numbers here as an 00:16:31.870 --> 00:16:33.909 approximation i think that the things that are 00:16:33.909 --> 00:16:36.090 important to note are what is the shape of the 00:16:36.090 --> 00:16:38.669 relationship in terms of like in the case of 00:16:38.669 --> 00:16:40.519 i guess kind of spoiler alert in the case of 00:16:40.519 --> 00:16:42.580 all of the lifts we found that like increasing 00:16:42.580 --> 00:16:45.700 the amount you jump up into a point will increase 00:16:45.700 --> 00:16:48.000 the probability of your lift being successful 00:16:48.000 --> 00:16:51.039 but there is an inflection point where that begins 00:16:51.039 --> 00:16:53.720 to change you know and and your probability of 00:16:53.720 --> 00:16:56.480 lifting so in in other words you can jump too 00:16:56.480 --> 00:16:58.799 much which i obviously is i think something that's 00:16:58.799 --> 00:17:00.879 not going to blow anybody's mind but the shape 00:17:00.879 --> 00:17:03.559 of the relationship is approximately similar 00:17:03.559 --> 00:17:07.279 across lifts but the the ranges are different 00:17:07.279 --> 00:17:09.759 which again so you need to know things like the 00:17:09.759 --> 00:17:12.339 range with which I can jump between squat attempts 00:17:12.339 --> 00:17:15.859 is larger than the range within which I can jump 00:17:15.859 --> 00:17:18.319 safely between bench press attempts. When it 00:17:18.319 --> 00:17:21.240 comes to individual lifters, then I think that 00:17:21.240 --> 00:17:23.519 is probably the biggest part where like knowing 00:17:23.519 --> 00:17:25.559 your athletes comes in because like, I mean, 00:17:25.579 --> 00:17:28.640 anyone who's been around lifters before, I knew 00:17:28.640 --> 00:17:31.180 people, people of similar weight classes to me 00:17:31.180 --> 00:17:33.720 or maybe marginally heavier who could make enormous 00:17:33.720 --> 00:17:35.940 jumps that I just, you know what I mean? Like 00:17:35.940 --> 00:17:39.660 for me, if I was doing squats, say, like anything 00:17:39.660 --> 00:17:43.039 more than 10 kilos would be very, very considerable 00:17:43.039 --> 00:17:45.539 for me to jump, even from warmup weights. I just, 00:17:45.660 --> 00:17:48.480 you know, it, it, it messed with the way my squats 00:17:48.480 --> 00:17:50.799 felt and, you know, my, I would misgrew things 00:17:50.799 --> 00:17:53.299 and stuff like that. So I had to be very meticulous 00:17:53.299 --> 00:17:55.859 in my warmups for squatting. Whereas I knew people 00:17:55.859 --> 00:17:58.700 who could come in 70 kilos on the bar, two reps, 00:17:58.920 --> 00:18:02.480 140 kilos on the bar, two reps, 180 kilos on 00:18:02.480 --> 00:18:04.619 the bar, five sets of eight reps. And he was 00:18:04.619 --> 00:18:06.740 just like, right. Yeah, I can't, I can't do that. 00:18:06.779 --> 00:18:08.630 You know, that kind of work. Or, you know, like 00:18:08.630 --> 00:18:11.009 getting to their work sets as quickly as possible, 00:18:11.130 --> 00:18:13.190 you know. So and, you know, those might be and 00:18:13.190 --> 00:18:14.990 you might see that across lifters with very similar 00:18:14.990 --> 00:18:17.569 profiles in terms of similar one rep maxes, so 00:18:17.569 --> 00:18:19.529 on and so forth, similar body weights. So that 00:18:19.529 --> 00:18:21.769 is obviously where the nuance of that coming 00:18:21.769 --> 00:18:23.470 in and knowing that that relationship will be 00:18:23.470 --> 00:18:26.269 consistent across lifters. But like the ranges 00:18:26.269 --> 00:18:29.589 might vary by lifter by lifter. And some of the 00:18:29.589 --> 00:18:33.309 factors we explore is meant to assist in understanding 00:18:33.309 --> 00:18:36.569 factors that may influence those ranges, essentially. 00:18:37.069 --> 00:18:39.049 I know we've probably kind of jumped ahead a 00:18:39.049 --> 00:18:41.170 little bit by talking about that, but yeah, that 00:18:41.170 --> 00:18:43.369 is in essence kind of what we were trying to 00:18:43.369 --> 00:18:45.089 do. And I think we've accomplished it to some 00:18:45.089 --> 00:18:47.849 extent. I guess the very boring result that I 00:18:47.849 --> 00:18:50.029 will mention at the beginning is that our model, 00:18:50.089 --> 00:18:52.170 so we separated again, just to reemphasize, we 00:18:52.170 --> 00:18:54.509 did a male and a female model. And again, that 00:18:54.509 --> 00:18:56.430 was partially for interpretability and partially 00:18:56.430 --> 00:18:59.069 just to avoid the male data overly influencing 00:18:59.069 --> 00:19:01.309 the female data from a predictive standpoint. 00:19:01.609 --> 00:19:04.569 But our two models had, there's a model parameter, 00:19:04.670 --> 00:19:06.210 it's the only one I'll talk about, which is called... 00:19:06.279 --> 00:19:08.259 sensitivity, which is essentially an indication 00:19:08.259 --> 00:19:11.960 of how good the model is at predicting a successful 00:19:11.960 --> 00:19:15.200 lift attempt. And they had a sensitivity for 00:19:15.200 --> 00:19:17.599 the male model. It was 87 % and the female model, 00:19:17.640 --> 00:19:20.160 I think it was 88%. So that's, you know, if you're 00:19:20.160 --> 00:19:21.880 rounding it up, I'll be, I'll be nice to myself 00:19:21.880 --> 00:19:25.150 and round it up. Nine out of 10. successful lifts 00:19:25.150 --> 00:19:28.190 in both models were detected by the model correct 00:19:28.190 --> 00:19:30.109 were predicted by the model correctly you know 00:19:30.109 --> 00:19:32.670 what i mean so nine out of ten times if you make 00:19:32.670 --> 00:19:34.789 a lift this model will will have predicted that 00:19:34.789 --> 00:19:36.750 you were going to make it you know whether you're 00:19:36.750 --> 00:19:38.730 a male or female before we kind of maybe talk 00:19:38.730 --> 00:19:41.490 about the jump um effect size stuff i think there's 00:19:41.490 --> 00:19:44.470 there's two things on the descriptive side that 00:19:44.470 --> 00:19:47.009 i thought when we were discussing the results 00:19:47.009 --> 00:19:48.849 as a research team we were kind of working together 00:19:48.849 --> 00:19:51.589 on it i thought were very very interesting some 00:19:51.589 --> 00:19:53.799 of my colleagues didn't so much but i I thought 00:19:53.799 --> 00:19:55.200 they were useful to go into. The first thing 00:19:55.200 --> 00:19:57.380 we did is we did on the descriptive level, which 00:19:57.380 --> 00:20:00.359 lifts had the highest overall success rates. 00:20:00.380 --> 00:20:02.700 So squat bench and deadlift for males and females. 00:20:02.880 --> 00:20:04.859 They're slightly different between sexes, but 00:20:04.859 --> 00:20:06.720 I thought it would just be interesting to note. 00:20:06.839 --> 00:20:09.539 So I think for both males and females, the bench 00:20:09.539 --> 00:20:13.200 press and the bench press had the lowest success 00:20:13.200 --> 00:20:17.640 rate. So I think it's around low 60%, like 62 00:20:17.640 --> 00:20:20.819 to 65 kind of percent, give or take for males 00:20:20.819 --> 00:20:22.339 and females. I think it was slightly lower for 00:20:22.339 --> 00:20:25.079 females. So irrespective of attempt or anything, 00:20:25.259 --> 00:20:27.960 you've only got about a 60 to 65 % chance of 00:20:27.960 --> 00:20:30.339 making a bench press lift. In comparison, the 00:20:30.339 --> 00:20:32.420 squats and the deadlifts were both in the mid 00:20:32.420 --> 00:20:36.420 to high 70%. So people tend to, you know, miss. 00:20:36.559 --> 00:20:39.259 The rate of bench press success is about 10 to 00:20:39.259 --> 00:20:42.539 12 % lower for both males and females and then 00:20:42.539 --> 00:20:45.519 either of the other lifts. And it's something 00:20:45.519 --> 00:20:47.859 I think that comes across consistently throughout 00:20:47.859 --> 00:20:49.859 our results is that the bench press is a bit 00:20:49.859 --> 00:20:52.079 of a bizarre animal in terms of some... some 00:20:52.079 --> 00:20:54.619 of the results we see. But I think that is anecdotally 00:20:54.619 --> 00:20:56.700 known in the powerlifting community. I think 00:20:56.700 --> 00:20:58.900 there's a term people use, poverty bench, which 00:20:58.900 --> 00:21:01.160 is basically you've got someone who's a good 00:21:01.160 --> 00:21:02.960 squatter and a good deadlifter, but just for 00:21:02.960 --> 00:21:04.779 whatever reason, an absolutely terrible bench 00:21:04.779 --> 00:21:07.380 presser. And, you know, I squat 250 kilos and 00:21:07.380 --> 00:21:10.279 I can't bench 110 kilos, you know, kind of situation. 00:21:10.539 --> 00:21:12.759 But I thought that was interesting. The other 00:21:12.759 --> 00:21:15.460 thing that I think is worth commenting on is 00:21:15.460 --> 00:21:18.319 that the irrespective of anything else, what 00:21:18.319 --> 00:21:20.240 lift you're doing, your age, anything like that, 00:21:20.319 --> 00:21:23.109 the probability of a successful third attempt 00:21:23.109 --> 00:21:28.329 was around 10 to 12 % lower than the probability 00:21:28.329 --> 00:21:31.650 of a successful second attempt. So all of the 00:21:31.650 --> 00:21:35.009 situations being equal, people tend to miss more 00:21:35.009 --> 00:21:36.730 on third attempts with that. The probability 00:21:36.730 --> 00:21:39.509 of a successful third attempt is around 50%. 00:21:39.509 --> 00:21:42.109 And the probability of a second attempt is a 00:21:42.109 --> 00:21:45.670 low 62, 63%, you know, kind of. And those numbers 00:21:45.670 --> 00:21:48.250 will vary marginally between males and females, 00:21:48.390 --> 00:21:51.190 but generally speaking. All other factors being 00:21:51.190 --> 00:21:53.690 equal, you know, you have a one in two chance 00:21:53.690 --> 00:21:56.269 of making a third attempt in comparison to a 00:21:56.269 --> 00:21:58.430 second attempt. I thought that was interesting. 00:21:58.490 --> 00:22:00.950 Now, and there is some other literature, might 00:22:00.950 --> 00:22:03.509 be a study by Kyle Travis. I think it might be 00:22:03.509 --> 00:22:06.130 2019 study. We cited it within our article, but 00:22:06.130 --> 00:22:08.309 they looked at, if I remember rightly, I think 00:22:08.309 --> 00:22:10.430 they looked at world championship, like IPF world 00:22:10.430 --> 00:22:12.730 championship play, I think top 10 or something 00:22:12.730 --> 00:22:15.170 like that. And they did estimate that you will 00:22:15.170 --> 00:22:17.430 see from first attempt to second attempt to third 00:22:17.430 --> 00:22:19.920 attempt, like a progressive decrease. in the 00:22:19.920 --> 00:22:22.319 rates of a successful lift across all three lifts. 00:22:22.420 --> 00:22:24.819 But I might be misremembering the authors of 00:22:24.819 --> 00:22:26.799 that study, but it is cited in our article. I 00:22:26.799 --> 00:22:28.880 just thought it was interesting. There are factors 00:22:28.880 --> 00:22:31.240 at play there in the sense that obviously people 00:22:31.240 --> 00:22:33.980 might take chances on their third attempt more 00:22:33.980 --> 00:22:36.579 and therefore, you know, are more likely to miss. 00:22:36.660 --> 00:22:38.579 But it is, I think that is something that's useful 00:22:38.579 --> 00:22:42.240 to know for coaches in that, like all other factors 00:22:42.240 --> 00:22:44.279 being equal before you start thinking about how 00:22:44.279 --> 00:22:46.559 you're handling your lifter and, you know, planning 00:22:46.559 --> 00:22:48.779 their lifting day strategy. There is kind of... 00:22:48.779 --> 00:22:52.319 of a natural progressive decline in success rates 00:22:52.319 --> 00:22:54.319 across attempts. And that's useful to know because 00:22:54.319 --> 00:22:56.799 I think it is useful for informing. It is information 00:22:56.799 --> 00:22:58.960 that can be used to inform attempt strategies, 00:22:59.160 --> 00:23:00.920 you know, but that's all just some descriptive 00:23:00.920 --> 00:23:03.779 stuff that we found before kind of getting into 00:23:03.779 --> 00:23:06.839 the nitty gritty of the model that I personally 00:23:06.839 --> 00:23:08.720 thought was actually useful findings from the 00:23:08.720 --> 00:23:10.980 study, but it could just be me. No, it makes 00:23:10.980 --> 00:23:13.480 sense. If coaches are coaching for a long time, 00:23:13.539 --> 00:23:16.160 they have this feeling that whatever jumps you 00:23:16.160 --> 00:23:18.920 can make between the first and second. the second 00:23:18.920 --> 00:23:21.920 and third the jump should be smaller or that 00:23:21.920 --> 00:23:25.359 you have to accept that the probability for lifter 00:23:25.359 --> 00:23:28.740 to make this lift happen will be lower yeah and 00:23:28.740 --> 00:23:31.759 it's potentially something where if you're going 00:23:31.759 --> 00:23:34.599 to make an ambitious jump it might be worth doing 00:23:34.599 --> 00:23:37.339 it after your first attempt but before your third 00:23:37.339 --> 00:23:38.799 attempt you know and I think that's something 00:23:38.799 --> 00:23:41.420 that people might intuitively do anyway particularly 00:23:41.420 --> 00:23:44.279 when I was competing but we had like there was 00:23:44.279 --> 00:23:46.079 a lot of these phrases that got kind of thrown 00:23:46.079 --> 00:23:48.029 around but I think it was like first attempt 00:23:48.029 --> 00:23:50.190 is to stay in it so you don't bomb out second 00:23:50.190 --> 00:23:52.630 attempt is to is to win it and then the third 00:23:52.630 --> 00:23:54.450 attempt is kind of in your back pocket if you 00:23:54.450 --> 00:23:56.990 need it you know so like whatever your best score 00:23:56.990 --> 00:23:59.309 of the day you thought was going to be try to 00:23:59.309 --> 00:24:01.109 get it on your second attempt so that you know 00:24:01.109 --> 00:24:03.099 you you have a conservative opener You're going 00:24:03.099 --> 00:24:04.720 to stay in the competition. You're going to register 00:24:04.720 --> 00:24:06.900 a total. Take your chance on your second attempt. 00:24:07.160 --> 00:24:09.119 If you miss it, you've got another chance at 00:24:09.119 --> 00:24:11.359 it. Great. If you make it, hey, there's a bit 00:24:11.359 --> 00:24:13.660 of wiggle room here. You know, we can get a few 00:24:13.660 --> 00:24:16.619 extra kilos on the board for placing, PBs, whatever, 00:24:16.839 --> 00:24:19.240 you know, that sort of stuff. And I guess I think 00:24:19.240 --> 00:24:22.299 that kind of that anecdotal strategy does also 00:24:22.299 --> 00:24:24.779 seem to line up a little bit with what we've 00:24:24.779 --> 00:24:27.019 asked, like with that result from our study. 00:24:27.200 --> 00:24:30.240 So in some sense, probably validating what at 00:24:30.240 --> 00:24:32.759 least some coaches are doing, but also. helps 00:24:32.759 --> 00:24:34.920 to emphasize that point and that you can quantify 00:24:34.920 --> 00:24:36.880 it with numbers to say that this is a real thing. 00:24:36.980 --> 00:24:40.599 You know, it helps, maybe helps emphasize that 00:24:40.599 --> 00:24:43.119 it is important to consider those sort of factors 00:24:43.119 --> 00:24:45.599 when it comes to planning competition strategies. 00:24:46.059 --> 00:24:48.539 Maybe, I guess, we could talk about the jump 00:24:48.539 --> 00:24:51.259 size effects, which are the main aspect of the 00:24:51.259 --> 00:24:54.200 model. So I guess kind of talking about it, I 00:24:54.200 --> 00:24:56.019 guess we have to kind of partition this by sex 00:24:56.019 --> 00:24:58.420 a little bit. But what we found, generally speaking 00:24:58.420 --> 00:25:01.220 for both sexes, without some minor, minor detail, 00:25:01.420 --> 00:25:04.019 generally the squat and the deadlift kind of 00:25:04.019 --> 00:25:07.279 responded the same in terms of jump attempt size 00:25:07.279 --> 00:25:10.019 in the sense that for males for example you can 00:25:10.019 --> 00:25:14.099 see like anywhere between 5 to 20 kilograms on 00:25:14.099 --> 00:25:16.420 both the squat and the deadlift you will actually 00:25:16.420 --> 00:25:19.819 see an increase in your probability of a successful 00:25:19.819 --> 00:25:22.720 attempt if you jump anywhere between 5 and anywhere 00:25:22.720 --> 00:25:25.599 under 20 and again that's going to be where you 00:25:25.599 --> 00:25:28.039 know and I think those are quite reasonable jumps 00:25:28.039 --> 00:25:30.839 that people tend to make on those lifts And obviously 00:25:30.839 --> 00:25:33.359 the individual variability is what decides because 00:25:33.359 --> 00:25:35.140 there's still quite a big range between five 00:25:35.140 --> 00:25:37.880 and 20 kilos. But that's where individual circumstances 00:25:37.880 --> 00:25:40.680 kind of come into play in that middle zone. Generally 00:25:40.680 --> 00:25:43.140 speaking, you will see declines in the probability 00:25:43.140 --> 00:25:46.140 of success in squats or deadlifts if you are 00:25:46.140 --> 00:25:48.480 a male with jumps over 20 kilos. And that effect 00:25:48.480 --> 00:25:50.720 was a little bit stronger on third attempts in 00:25:50.720 --> 00:25:52.759 comparison to second attempts, which I think 00:25:52.759 --> 00:25:55.359 is, you know, by the time you reach your third 00:25:55.359 --> 00:25:57.440 deadlift, you're pretty tired. Jumping 20 kilos 00:25:57.440 --> 00:26:00.210 is a tall order. For males on the... Bench press, 00:26:00.410 --> 00:26:03.329 jumping less than 10 kilos, generally speaking, 00:26:03.549 --> 00:26:06.289 between attempts will improve or you'll have 00:26:06.289 --> 00:26:08.589 higher success if you jump less than 10 kilos. 00:26:08.690 --> 00:26:11.650 But you will see that like there's a really sharp 00:26:11.650 --> 00:26:14.410 drop off with larger jumps on the third attempt 00:26:14.410 --> 00:26:16.430 again in comparison to the second attempt. And 00:26:16.430 --> 00:26:19.210 that's really just a more granular manifestation 00:26:19.210 --> 00:26:21.329 of what I was talking about previously when we 00:26:21.329 --> 00:26:23.529 talked about third attempt success rates in comparison 00:26:23.529 --> 00:26:26.190 to second attempt success rates. For females, 00:26:26.450 --> 00:26:29.329 you'll see kind of something similar. on the 00:26:29.329 --> 00:26:31.970 squat and deadlift in that like you can see, 00:26:31.970 --> 00:26:35.109 you know, increasing success rates, but the range 00:26:35.109 --> 00:26:37.509 is smaller. So generally for females based on 00:26:37.509 --> 00:26:40.809 our model between nine and 11 kilos or eight 00:26:40.809 --> 00:26:43.289 and 11 kilos in and around is where you will 00:26:43.289 --> 00:26:46.089 see kind of you're likely to see an increased 00:26:46.089 --> 00:26:49.170 probability of success in comparison to males 00:26:49.170 --> 00:26:51.450 who were five to 20. So that range is tighter, 00:26:51.529 --> 00:26:53.670 but the principle is the same. Something we saw 00:26:53.670 --> 00:26:55.670 with females. And again, this is probably something 00:26:55.670 --> 00:26:57.450 that you would see a little bit of an anecdote 00:26:57.450 --> 00:27:00.619 in that. like the extent to which females see 00:27:00.619 --> 00:27:03.819 an increase in the probability of bench press 00:27:03.819 --> 00:27:06.460 success is very, very small. So I think we said 00:27:06.460 --> 00:27:08.980 anywhere, it's like something like between two 00:27:08.980 --> 00:27:11.059 and four kilos or something like that as estimated 00:27:11.059 --> 00:27:13.980 by the model. And I mean, yeah. And again, I 00:27:13.980 --> 00:27:16.039 think we saw like a marginally lower probability 00:27:16.039 --> 00:27:18.299 of success rate on the bench press for females 00:27:18.299 --> 00:27:21.279 as well. And particularly on third attempt size, 00:27:21.400 --> 00:27:24.400 like any kind of moderate, you'll see it in one 00:27:24.400 --> 00:27:26.819 of the figures in the study. If I was to pull 00:27:26.819 --> 00:27:28.359 it up just for any. who wants to kind of read 00:27:28.359 --> 00:27:30.359 the study, all two of the people who want to 00:27:30.359 --> 00:27:34.099 read this study, in figure 4C on the right -hand 00:27:34.099 --> 00:27:37.099 side, you'll see kind of the female bench press 00:27:37.099 --> 00:27:39.720 result is like a complete, it's like a waterfall. 00:27:39.880 --> 00:27:41.500 You know what I mean? It comes up within a very 00:27:41.500 --> 00:27:43.759 narrow range and then drops off very, very hard. 00:27:43.920 --> 00:27:46.640 It's difficult. You know, it is in theory a similar 00:27:46.640 --> 00:27:49.380 relationship to what the males see, but it's 00:27:49.380 --> 00:27:52.619 considerably more pronounced. And it's difficult 00:27:52.619 --> 00:27:56.619 to know why that is. There's, I mean, it's consistent. 00:27:56.720 --> 00:27:59.619 in the literature that females display lower 00:27:59.619 --> 00:28:02.700 estimates of upper body strength relative to 00:28:02.700 --> 00:28:05.099 males, even in comparison to like lower body 00:28:05.099 --> 00:28:08.460 strength differences. I think Tommy Lundberg 00:28:08.460 --> 00:28:10.539 from the Karolinska Institute has a really nice 00:28:10.539 --> 00:28:12.680 large scale study where they just compared different 00:28:12.680 --> 00:28:15.180 athletic kind of traits between males and females. 00:28:15.440 --> 00:28:17.420 I think you see things like very large differences 00:28:17.420 --> 00:28:19.539 between males and females in terms of things 00:28:19.539 --> 00:28:21.460 like punching power and so on and so forth. So 00:28:21.460 --> 00:28:23.460 I only hide that to say that that difference 00:28:23.460 --> 00:28:26.650 is pronounced across other like upper. body strength 00:28:26.650 --> 00:28:29.049 and power related kind of movements between males 00:28:29.049 --> 00:28:31.829 and females. But I actually, from what I understand, 00:28:32.029 --> 00:28:35.250 the mechanisms behind that are not well understood. 00:28:35.529 --> 00:28:37.210 I think there's some stuff around muscle mass 00:28:37.210 --> 00:28:39.329 distribution and stuff like that, but I find 00:28:39.329 --> 00:28:41.049 it difficult to believe that that is the whole 00:28:41.049 --> 00:28:43.250 story. But it is something that's identified 00:28:43.250 --> 00:28:46.069 in our study as well. It is in somewhat, in some 00:28:46.069 --> 00:28:49.089 ways consistent with other similar athletic style 00:28:49.089 --> 00:28:52.650 events, but it is difficult to understand why 00:28:52.650 --> 00:28:55.930 that relationship is much more profound. in females 00:28:55.930 --> 00:28:58.750 compared to males, but it is there. So I guess 00:28:58.750 --> 00:29:01.710 kind of wrapping that up, that section of it, 00:29:01.789 --> 00:29:04.309 it's to say that squats and deadlifts, generally 00:29:04.309 --> 00:29:06.230 speaking, have similar relationships for both 00:29:06.230 --> 00:29:08.650 males and females in the sense that larger jumps 00:29:08.650 --> 00:29:11.390 can be tolerated and larger jumps will, within 00:29:11.390 --> 00:29:14.549 certain ranges, increase the probability of successful 00:29:14.549 --> 00:29:17.109 lifts. So you can actually, to some extent, jump 00:29:17.109 --> 00:29:19.190 too small. Like when we talk about the upper 00:29:19.190 --> 00:29:21.450 limits of the jump size in this study, like, 00:29:21.470 --> 00:29:24.150 you know, I think the upper limits are more represented. 00:29:24.359 --> 00:29:26.059 of a physical limits there and that like you 00:29:26.059 --> 00:29:28.460 can literally only lift so much. Like I'm not 00:29:28.460 --> 00:29:30.559 gonna, I don't think anyone ever missed a 25 00:29:30.559 --> 00:29:32.640 kilo jump because they just didn't believe hard 00:29:32.640 --> 00:29:34.200 enough. You know, that kind of thing. It's probably 00:29:34.200 --> 00:29:37.000 just 25 kilos is really, really heavy. But when 00:29:37.000 --> 00:29:39.740 we talk more within those nuanced small ranges, 00:29:39.799 --> 00:29:42.039 like the difference between, let's say on a squat 00:29:42.039 --> 00:29:43.720 or a deadlift, the difference between making 00:29:43.720 --> 00:29:45.859 a two and a half kilo jump on a squat versus 00:29:45.859 --> 00:29:48.400 like eight kilo jump, six kilo jump, five kilo 00:29:48.400 --> 00:29:50.720 jump, something like that. That is probably where 00:29:50.720 --> 00:29:53.359 that kind of comes into play. And again, like. 00:29:53.599 --> 00:29:56.799 That is purely speculation by me. I don't fully 00:29:56.799 --> 00:29:58.740 know, but I think that is something that could 00:29:58.740 --> 00:30:01.500 be a factor of just like giving people an appropriate 00:30:01.500 --> 00:30:04.559 challenge. That would be my guess. Yeah. And 00:30:04.559 --> 00:30:06.539 it's probably a good time to point out in that, 00:30:06.640 --> 00:30:08.680 like, I think because you see this, I mean, it's 00:30:08.680 --> 00:30:10.579 a slight detraction from us in terms of we're 00:30:10.579 --> 00:30:12.640 building a model that's like more around competition 00:30:12.640 --> 00:30:14.920 strategy, but like, you know, a machine can build 00:30:14.920 --> 00:30:16.519 you a decent training program, but a machine 00:30:16.519 --> 00:30:18.200 can never coach you. You know what I mean? Like 00:30:18.200 --> 00:30:19.980 you see these AI apps and stuff like that, that 00:30:19.980 --> 00:30:21.940 people use. And I think the programs that they 00:30:21.940 --> 00:30:23.670 build on some of the better. ones are like perfectly 00:30:23.670 --> 00:30:26.009 fine you know what i mean but you know you're 00:30:26.009 --> 00:30:28.009 never you're not going to get like it's not going 00:30:28.009 --> 00:30:29.690 to help you on a competition day it's not going 00:30:29.690 --> 00:30:31.450 to critique your technique it's not going to 00:30:31.450 --> 00:30:33.210 help you with your mindset and stuff like that 00:30:33.210 --> 00:30:34.950 as well when it comes to approaching competition 00:30:34.950 --> 00:30:37.750 and i think that that is something very similar 00:30:37.750 --> 00:30:39.890 that i would highlight with respect to our model 00:30:39.890 --> 00:30:41.789 in the sense that this cop this this will never 00:30:41.789 --> 00:30:44.369 plan a competition strategy for you it will just 00:30:44.369 --> 00:30:47.230 maybe give you some information that is useful 00:30:47.230 --> 00:30:49.529 in your coach's bucket alongside everything else 00:30:49.529 --> 00:30:52.539 to help you make the application of the of coaching 00:30:52.539 --> 00:30:54.579 a little more informed and perhaps a little more 00:30:54.579 --> 00:30:57.079 systematic if possible. You know, that's kind 00:30:57.079 --> 00:30:59.519 of the intent behind it, at least. I think it's 00:30:59.519 --> 00:31:01.859 also useful to have you build models separately 00:31:01.859 --> 00:31:05.059 for females and males. And anecdotally, coaching 00:31:05.059 --> 00:31:08.279 females and males is slightly different. So if 00:31:08.279 --> 00:31:10.799 you can see it in the model too, then you can 00:31:10.799 --> 00:31:13.940 take it in and keep it in mind in terms of how 00:31:13.940 --> 00:31:17.119 do I even talk about the possible numbers? And 00:31:17.119 --> 00:31:20.000 if I have a lifter who is studying, how should 00:31:20.000 --> 00:31:23.759 I gauge this? standard way to coach the jumps 00:31:23.759 --> 00:31:27.019 in the warm up or in the approaching PR and so 00:31:27.019 --> 00:31:28.960 on because if you do things for the first time 00:31:28.960 --> 00:31:30.779 you don't have preferences you don't know so 00:31:30.779 --> 00:31:33.299 you have to start somewhere yeah and I mean this 00:31:33.299 --> 00:31:35.640 isn't something we looked at but something I 00:31:35.640 --> 00:31:37.759 did and it's more applicable again to weightlifting 00:31:37.759 --> 00:31:39.779 than powerlifting so we're really getting tangential 00:31:39.779 --> 00:31:41.759 now but one thing I noticed in weightlifting 00:31:41.759 --> 00:31:43.859 and a little bit I coach mostly males throughout 00:31:43.859 --> 00:31:46.099 my career just through happenstance I worked 00:31:46.099 --> 00:31:48.019 with some females but mostly males but something 00:31:48.019 --> 00:31:50.400 I did notice from like because the club I coach 00:31:50.480 --> 00:31:52.000 that was different from the club that I trained 00:31:52.000 --> 00:31:53.759 at. So some of the stuff I noticed with the girls 00:31:53.759 --> 00:31:56.240 that I trained with was they would always open 00:31:56.240 --> 00:31:58.660 the competitions like way closer to their max 00:31:58.660 --> 00:32:00.680 than I would. You know, like I would train with 00:32:00.680 --> 00:32:03.740 girls who might have like, say, 82, 83 kilos 00:32:03.740 --> 00:32:07.019 snatch and they'd be opening at like 78, 79 kilos. 00:32:07.319 --> 00:32:10.440 Whereas I might have like 125 kilos snatch. I'm 00:32:10.440 --> 00:32:12.759 going to open at like 115 kilos. You know, I 00:32:12.759 --> 00:32:14.799 mean, like way, way further away. But they were 00:32:14.799 --> 00:32:17.019 much more comfortable like opening really heavy 00:32:17.019 --> 00:32:19.119 and then making those more precise jumps where 00:32:19.119 --> 00:32:21.569 I was. like I need to open really light and make 00:32:21.569 --> 00:32:23.670 bigger jumps you know and that was something 00:32:23.670 --> 00:32:26.029 that I mean again it's pure anecdote and it's 00:32:26.029 --> 00:32:27.869 weightlifting instead of powerlifting but that 00:32:27.869 --> 00:32:30.269 was that was one of the factors that did inform 00:32:30.269 --> 00:32:33.049 separating out models that those models was just 00:32:33.049 --> 00:32:35.130 me being like well I have this experience of 00:32:35.130 --> 00:32:36.990 like whether it's physiological or psychological 00:32:36.990 --> 00:32:39.609 it's like just seeing girls approach it differently 00:32:39.609 --> 00:32:42.230 to how guys do you know and being like okay that 00:32:42.230 --> 00:32:44.470 you know you because whether it is physiological 00:32:44.470 --> 00:32:46.750 or psychological it may still manifest in the 00:32:46.750 --> 00:32:49.210 data if it's something that's kind of very consistent 00:32:49.210 --> 00:32:52.349 obviously that's pure anecdote but but i i do 00:32:52.349 --> 00:32:54.890 i do think there is something to it you know 00:32:54.890 --> 00:32:57.769 and to some degree you know we did some linear 00:32:57.769 --> 00:33:00.529 fixed effects and essentially again fancy statistical 00:33:00.529 --> 00:33:03.710 terms we included some things in the model that 00:33:03.710 --> 00:33:06.150 we did not make non -linear just because again 00:33:06.150 --> 00:33:08.029 just makes things more complicated makes things 00:33:08.029 --> 00:33:10.549 harder to interpret so we were like we'll just 00:33:10.549 --> 00:33:12.269 keep these as linear because it's much easier 00:33:12.269 --> 00:33:14.369 it's it's a much more straightforward interpretation 00:33:14.369 --> 00:33:18.400 but the caveat there is that the like take it 00:33:18.400 --> 00:33:20.319 with more of a grain of salt. Essentially, it's 00:33:20.319 --> 00:33:22.740 not as a precise finding as some of the stuff 00:33:22.740 --> 00:33:24.519 we've talked about already. So the three things 00:33:24.519 --> 00:33:27.039 we included in there were the first attempt weight. 00:33:27.099 --> 00:33:29.440 So the opening attempt weight, essentially. So 00:33:29.440 --> 00:33:31.740 what your opener is, body weight, your exact 00:33:31.740 --> 00:33:33.960 body weight, you know, as opposed to your weight 00:33:33.960 --> 00:33:36.160 class and age. And we included those as linear 00:33:36.160 --> 00:33:38.700 effects just to see how they also interacted 00:33:38.700 --> 00:33:41.319 with the probability of a successful attempt. 00:33:41.619 --> 00:33:44.200 We found opening weight had a negative association 00:33:44.200 --> 00:33:47.380 with the probability of lift success, which I 00:33:47.380 --> 00:33:49.500 guess, kind of make heavier you open, the less 00:33:49.500 --> 00:33:52.019 likely you are to make subsequent lifts. So again, 00:33:52.079 --> 00:33:53.940 there's probably a sweet spot there, but that's 00:33:53.940 --> 00:33:56.259 potentially not surprising. Body weight had a 00:33:56.259 --> 00:33:58.839 positive association with lift probability success. 00:33:59.000 --> 00:34:02.079 So heavier lifters tend to make more lifts. Again, 00:34:02.259 --> 00:34:04.920 that's within reason, probably something that 00:34:04.920 --> 00:34:06.859 makes sense. And likely these relationships, 00:34:07.039 --> 00:34:09.739 if we were to do very granulated analysis of 00:34:09.739 --> 00:34:13.039 them, would have some kind of a nonlinear relationship, 00:34:13.219 --> 00:34:15.659 but how I would interpret them within the context 00:34:15.659 --> 00:34:18.320 of... lifters here would be like, it's probably 00:34:18.320 --> 00:34:20.619 a good idea to fill out your weight class. So 00:34:20.619 --> 00:34:23.900 what I mean by that is if I lift in the 105 kilo 00:34:23.900 --> 00:34:26.579 weight class, it's not really valuable for me 00:34:26.579 --> 00:34:29.460 to be 100 kilos. I would probably have a better 00:34:29.460 --> 00:34:32.199 chance of making lifts if I weigh 106 kilos, 00:34:32.420 --> 00:34:34.840 don't have breakfast, weigh in at 105, happy 00:34:34.840 --> 00:34:37.000 days, you know, that kind of situation. So that 00:34:37.000 --> 00:34:38.659 would be the way I would be interpreting those 00:34:38.659 --> 00:34:41.199 relationships for now. Similarly, you know, it's 00:34:41.199 --> 00:34:43.039 useful to probably have a conservative goal putting 00:34:43.039 --> 00:34:46.170 weight. And very interestingly, we didn't. find 00:34:46.170 --> 00:34:49.369 that age meaningfully influenced successful lift 00:34:49.369 --> 00:34:51.389 probability I actually kind of thought you would 00:34:51.389 --> 00:34:53.610 either see something on either end of the spectrum 00:34:53.610 --> 00:34:56.590 where younger lifters would miss more lifts because 00:34:56.590 --> 00:34:59.429 they're perhaps overly enthusiastic or older 00:34:59.429 --> 00:35:01.630 lifters would miss more lifts because they're 00:35:01.630 --> 00:35:04.909 old you know but if you look at that figure in 00:35:04.909 --> 00:35:07.010 the study I think it might be figure five it's 00:35:07.010 --> 00:35:09.489 essentially like a horizontal line so there really 00:35:09.489 --> 00:35:12.190 wasn't a relationship there now that could also 00:35:12.190 --> 00:35:15.309 just say that even if younger lifters are more, 00:35:15.309 --> 00:35:18.130 and being anecdotal here, say chaotic in their 00:35:18.130 --> 00:35:20.389 lifting attempt strategies, they might be young 00:35:20.389 --> 00:35:22.530 enough to do that. Whereas older lifters may 00:35:22.530 --> 00:35:25.269 kind of learn to become conservative as they 00:35:25.269 --> 00:35:27.329 get older. And therefore, you know, that might 00:35:27.329 --> 00:35:30.349 be why that happens. Something that's probably 00:35:30.349 --> 00:35:32.829 very late in the game to mention, we did account 00:35:32.829 --> 00:35:36.050 for individual lifter variability in our model. 00:35:36.110 --> 00:35:39.570 So all that means is essentially, to some extent, 00:35:39.650 --> 00:35:42.289 we tried to capture the fact that like, if I'm 00:35:42.289 --> 00:35:44.050 in open powerlifting and I have three competitions, 00:35:44.199 --> 00:35:46.820 there's a relationship between those three competitions 00:35:46.820 --> 00:35:49.239 and there is a relationship between the attempt 00:35:49.239 --> 00:35:52.380 strategies that were applied by me within those 00:35:52.380 --> 00:35:54.860 competitions so I might be different from people 00:35:54.860 --> 00:35:58.019 who through all of the circumstances look similar 00:35:58.019 --> 00:36:00.480 to me you know in terms of guys who similar height 00:36:00.480 --> 00:36:02.539 similar weight similar lift numbers all that 00:36:02.539 --> 00:36:04.079 sort of stuff there's still going to be differences 00:36:04.079 --> 00:36:06.460 and I think we estimated that effect to be somewhere 00:36:06.460 --> 00:36:09.880 between 12 to 14 percent and what what that kind 00:36:09.880 --> 00:36:12.980 of just means is that like not it doesn't strictly 00:36:12.980 --> 00:36:15.800 mean that like the values range that I've given 00:36:15.800 --> 00:36:19.440 you can be estimated to vary between 12 to 14%. 00:36:19.440 --> 00:36:22.480 Instead, it just simply says that like the individual 00:36:22.480 --> 00:36:25.340 lifter is accounting for like differences between 00:36:25.340 --> 00:36:28.349 lifters is accounting for like kind of. 12 % 00:36:28.349 --> 00:36:30.949 of the error in our model, essentially. So difficult 00:36:30.949 --> 00:36:33.690 to interpret, but it is accounted for, which 00:36:33.690 --> 00:36:35.809 again, just adds some precision to our estimates 00:36:35.809 --> 00:36:38.969 when we generalize it. We're knowing which competitions 00:36:38.969 --> 00:36:41.090 are paired to the same lifter over and over and 00:36:41.090 --> 00:36:43.849 over again. But I think those are the last findings 00:36:43.849 --> 00:36:45.929 of the study, at least. Awesome. Huge study. 00:36:46.070 --> 00:36:49.789 If anyone is brave enough, they should go and 00:36:49.789 --> 00:36:52.110 at least look at the figures because I think 00:36:52.110 --> 00:36:54.869 that's helpful. You have lots of beautiful figures 00:36:54.869 --> 00:36:58.090 there. So that is... something I would recommend 00:36:58.090 --> 00:37:00.690 now I have two questions to finish the first 00:37:00.690 --> 00:37:03.710 one is what is your favorite lift my favorite 00:37:03.710 --> 00:37:06.469 lift so when I was I'm retired from weightlifting 00:37:06.469 --> 00:37:08.349 now but when I was weightlifting my favorite 00:37:08.349 --> 00:37:10.329 lift was the split jerk believe it or not I was 00:37:10.329 --> 00:37:13.929 a bit of a peculiar athlete in that I was I very 00:37:13.929 --> 00:37:17.409 rarely missed jerks I would very often miss cleans 00:37:17.409 --> 00:37:20.949 I wasn't particularly strong so I generally knew 00:37:20.949 --> 00:37:23.769 if I completed a clean I was going to make the 00:37:23.769 --> 00:37:26.269 jerk so I was you know I would finish the clean 00:37:26.349 --> 00:37:28.469 very difficult clean I get to the top of it I'm 00:37:28.469 --> 00:37:30.369 like oh yeah here we go here's the fun part so 00:37:30.369 --> 00:37:32.309 I used to I know a lot of people like the split 00:37:32.309 --> 00:37:34.329 jerk but I really enjoyed it now it's probably 00:37:34.329 --> 00:37:36.110 something like a back squat or something like 00:37:36.110 --> 00:37:37.869 that you know it makes me feel like I'm still 00:37:37.869 --> 00:37:40.429 vaguely athletic but those are my if I can give 00:37:40.429 --> 00:37:42.230 you two answers those are my two answers split 00:37:42.230 --> 00:37:45.230 jerk and back squat that will work and last question 00:37:45.230 --> 00:37:47.849 is where people can find you if they want to 00:37:47.849 --> 00:37:51.070 look at your work or see what you're up to yeah 00:37:51.070 --> 00:37:54.610 so I'm on Research Case Ian Darrah D -A -R -R 00:37:54.610 --> 00:37:58.239 -A -G -H And I am on LinkedIn as well under the 00:37:58.239 --> 00:38:00.579 same name. I have an Instagram. Instagram is 00:38:00.579 --> 00:38:02.480 probably the best place to contact me if you're 00:38:02.480 --> 00:38:05.039 looking to, say, get a copy of the study or if 00:38:05.039 --> 00:38:07.619 you are looking to just reach out with a question 00:38:07.619 --> 00:38:09.179 or anything like that. That is the best place 00:38:09.179 --> 00:38:11.960 to find me. Bean Swole is my Instagram. Bean 00:38:11.960 --> 00:38:15.400 as in the legume and Swole as in I once was Swole. 00:38:15.539 --> 00:38:17.980 So if you're trying to find me there, that's 00:38:17.980 --> 00:38:19.559 probably the best place to contact me. Otherwise, 00:38:19.639 --> 00:38:21.699 LinkedIn or ResearchGate. I think the article 00:38:21.699 --> 00:38:23.860 is available for free on ResearchGate. If not, 00:38:23.900 --> 00:38:25.880 you can request it. then I'll provide it to you. 00:38:25.940 --> 00:38:28.260 If people are looking for full texts of the article, 00:38:28.380 --> 00:38:30.179 I'm happy to provide that. Awesome. Thank you 00:38:30.179 --> 00:38:32.460 so much, Ian. Pleasure. Thank you. Yeah, no, 00:38:32.539 --> 00:38:34.000 the pleasure was all mine. I really enjoyed it.