Where will sports scouting end?
This might sound silly: sports won’t end; therefore, scouting never will. But looking forward to the indefinite future, there are timelines where techniques converge, and development could plateau. I also present timelines (that I believe to be plausible) where techniques continue to increase in complexity and variety.
In this piece, I will use scouting to more broadly represent the whole player acquisition process, not just “people watching games”. You should note, therefore, that all references to scouting are referring to a concept and process much broader than the conventional role of a scout.
I will overview the key directions of development in the last few decades, for any less technical/nerdy readers, then lay out 4 hypothetical timelines for the future of sports scouting and player acquisition, as follows:
- “Information Maximisation Model”
- “Competitive Niches Model”
- “Unilateral Information Reduction Legislation”
- “Persistent Expansion Model”
The Modern Landscape
To understand why I have chosen these timelines, we first need a basic summary of the current landscape. Feel free to skip this section if you’re a ‘data person’.
The infamous ‘Moneyball’ tactics hit baseball 25 years ago, and teams started evaluating player price primarily with performance metrics. We haven’t, though, seen the death of conventional scouting in the following period. Some teams succeed well based on brilliant intuition, and coaches spotting development opportunities.
While traditional techniques live on, new frontiers are constantly being explored. The book ‘The MVP Machine’ sets out brilliantly the follow-on from Moneyball, where data-minded people turned their methods to player improvement. The crux of it is that now players can be statistically evaluated not just on who they are, but who they could become under the correct coaching regimen.
Teams like Brighton and Brentford in football have built huge outliers, with success far beyond what their budgets should ever offer. They were overwhelmingly not the first to do data in Football. What mattered for them was not getting a jump on the crowd like the Oakland A’s did, but being the best at understanding how to mesh together coaching, tactics, player roles and analytics. There are many great, promising young players that Brentford would never sign, not because they wouldn’t be good but because they don’t fit the role definitions and squad composition they were seeking.
Their millions were made on saying things like “we don’t want to find a good Right Back; we want to find the best crosser with good pace and work rate”. It’s like the difference between buying a Mercedes because it’s a really good car, and buying a Hilux because you specifically ranked all cars by the best for your use case: a 400km drive across the outback. Your Hilux is worse at loads of other things; it doesn’t even have heated seats. But you don’t need it to have heated seats.
I tell you this only to illustrate that, in the middle of all of this success for data-people of different kinds (of which there are too many to list), new techniques and creative implementations are on the rise. There has been a massive fanning out of technology, ideas, books and theories. Certainly it’s true that there’s never in history been more ways to measure a cricketer, and I’d imagine it’s the same for most other sports with player acquisition requirements. The pursuit of information is not just related to ability either.
Nowadays, football medical checks are highly advanced, and in cricket several stars have been rested due to stress fractures in the spine small enough that there’s no significant pain yet, but it’s indicative they are risking a big injury if they play. This isn’t out of goodwill to treat the players more caringly. It’s because 30 years ago your team would’ve signed the player with a stress fracture none-the-wiser, and everything would’ve been fine… until it wasn’t.
All of this matters because it’s more information. Many economic thinkers focus heavily on the role of information in markets. Akerlof’s ‘Market For Lemons’ thought experiment demonstrated that if there’s an information imbalance, people are driven away from selling their high-quality assets, and the market just becomes disproportionately full of duds. In sporting terms, this would be equivalent to listing your star striker with a known recurring knee problem and trying to falsely sell him off as a fit athlete in their prime. And inversely, if you’ve got someone with amazing fitness who will never get injured, why sell him if that bulletproofness isn’t reflected in the price tag?
The follow-on from this, therefore, is that fairer prices, and hence better deals, can be achieved by people with more information. If you know precisely the condition of someone’s body, you can more accurately estimate their sporting worth, and have a competitive advantage over those who are selecting teams without this knowledge.
And so today we land in a position of great knowledge; the teams without it have been significantly disadvantaged, make worse acquisitions, train their players with inferior plans and are outcompeted. That remains the case until they invest in catching up in the information race.
Scenario 1: The Information Maximisation Model
The knowledge direction discussed before drives us towards our first theory. As discussed, the current motion of scouting is increasing information and complexity. All sporting teams are looking to have more information, and thus make valuations more accurately. If we acknowledge there is only so much information that can be gleaned about an athlete, a scenario where there is an end point opens up. The theory is as follows:
Statement 1:
There is a finite amount of information that can be attained about an athlete.
Statement 2:
It is in every competing team’s interest to attain all of the information available.
Statement 3:
In elite sports, the budgets are high enough that there will be no methods which are only financially viable to some teams.
If all 3 statements are true:
We will reach a point where all teams are attaining as much information as possible, scouting with all the same methods, doing the same medical tests and using all the same data to predict player values. Any team not doing it will fall behind, so everyone will have to do it just to stay level. At this end point, there will not be a possible further step available, so scouting methods will stall.
As this stalled state is approached, people will find new previously untapped potentials, but they will have increasingly diminished impacts as the final possible morsels of extra knowledge are squeezed out. Knowing whether or not a player has a severe mental health condition could be worth millions, but perhaps knowing that the way they throw a ball is 2% more likely to cause an injury than optimal technique is worth no more than tens of thousands.
There is plenty of historical evidence for the processes this model describes occurring on a small scale. One such case, for example, is Baseball in the Dominican Republic. Once an untapped market, the Dodgers opened a fully Dodgers-owned academy there in 1987. Enjoying their own personal niche and picking up undervalued talent from their own private talent pool. By 2003, all 30 MLB teams had an academy in the Dominican Republic. Now everyone fishes from the Dominican pool. A once unique method has become ubiquitous.
In the first IPL cricket auction in 2008, one of the smallest teams is generally agreed to have ‘Moneyballed’ it and won the title with a team of undervalued domestic players and out-of-favour internationals. Nowadays every team is extensively strategising with statistics live at the auction table. I mention this example specifically because it’s important to understand that theories resulting in convergent scouting/information do not denote a convergence of team results. The uniformity of one aspect of contest simply increases the relative importance of others.
In football, these other factors may be managers or revenue imbalances (since spending limits are tied to income). In MLB this seems most likely to be purely budget, since there’s no salary cap. In the IPL, which has a partial cost cap, the results may be increasingly decided by which cities are producing the best local talent and who offers the most lucrative salaries to players retained with uncapped wages. This is shown by the long-term dominance of Chennai, Mumbai and Kolkata, who have won 13 of the 18 seasons. All three are immensely rich franchises with heavily ingrained cricket cultures and a history of producing talent and filling stadiums. I’m not suggesting that the IPL has reached this convergent end point by any means, but it does neatly highlight what sporting outcomes we would expect to see in a relatively uniform data market.
In broader terms, what this will look like long-term is a lack of new teams breaking through. In football, we would expect to only see new breakthrough sides of the Wrexham nature, where it is financial investment that drives the progress, rather than new clubs copying the routes of the previously mentioned Brighton and Brentford, who majorly outperformed their opponents in the transfer market.
When talking about franchised versus non-franchised sports, there is an important division we must stop to make. This IM Model assumes functionally infinite wealth compared to analytical expense. For a franchise like Mumbai Indians worth over a billion dollars, paying someone a day-rate to go and watch an amateur game for you, or buying a 20,000 USD camera rig, does effectively cost nothing, making this end-point feasible.
In sports with a pyramid, like English Football, the money must stop somewhere. It would be feasible for the top 12, or even the whole Premier League, to end up in this convergent model, but what then of the 3rd tier clubs, or the 5th tier clubs? There surely will never be unlimited resource funding there. Since Statement 3 is false, convergence in scouting through this scenario cannot occur. In that lower league, then you could out-scout your rivals, punch above your weight and get a promotion or perhaps two. But survival in this case at the top level becomes brutally tough because now you are fishing from a player pool in which every other fisherman has the best gear available. The bargains will be gone, and the survival of such a ‘data club’ or ‘scouting club’ will lie almost exclusively on their non-scouting attributes.
Scenario 2: Unilateral Information Reduction Model
This scenario only occurs after a case of the IM Model, or fear as Scenario 1 conditions are being approached. In the IM Model, we consider every team, let’s say in the MLB, to be doing expensive medical tests, expensive data collection, expensive in-person scouting, expensive psychometric tests (etc…). In this case, nobody has an advantage; the playing field is level in terms of recruitment, but everyone is spending a lot of money. In this situation, it becomes a serious option for consideration of some kind of non–compete agreement, banning certain methods or introducing a cost cap.
If teams were cost-capped tightly on recruitment, they would be forced to choose between medical tests and extensive scouting, or have to pick and choose which subset of baseball players they wanted to seriously scout/assess. By limiting this, divergence is created because it will no longer be optimal for everyone to allocate their budget to doing the same thing; doing something different to everyone else gives you an ‘untapped market’ of players, and there will be space to do something different, because nobody will be able to cover everything themselves.
Banning specific methods like advanced medical tests, psychometric tests or collecting certain types of data would do 2 things:
-
Reduce everyone’s expenses while maintaining a relatively balanced playing field.
-
Create two separate incentives to cheat:
2a) Follow an Akerlof model and just offer up your injured duds for trade, or people with a problem outside the allowed information, like a psychological issue with pressure situations or low ability to learn new concepts.
2b) Use the banned methods for a competitive advantage in recruitment, worth an immense amount financially if you are not caught.
If these 2 issues with cheating incentives could be mitigated, a methodological ban could benefit every team, saving them money and maintaining an even playing field. Doing such mitigation appears seriously difficult, though. Cost caps are generally easier to enforce, and the competitive context of sport may lead people to feel a cost-capped environment is more organic. With bans on methods, everyone must conform to artificial shared practices, but under a cost cap it remains theoretically possible to get an edge through creative techniques, so I suspect that approach would be preferred ideologically. Cost caps exist in a wide range of sporting environments to aim to prevent this unhelpful arms race of overspending, with mixed effects.
The real problem with cost cap solutions in this situation is the relative cheapness of data analytics. Teams likely would argue that good club doctors and health checks are essential for player safety, so those should be exempt. Then you consider that watching video footage is far cheaper than attending games, because people like me would love to spend hours running models on player attributes for really very modest sums of money. Many would (probably rightly) argue that it would be sad to see a cost cap destroy in-person scouting and replace it fully with offices and laptops.
I don’t think any of this is actually a successful put-down of cost cap models though, just demonstrations of pitfalls. A cap that had separate components for medical expenses, travelling scouts, data analysts, pitch labs, etc. could act to enforce a more modest, but still diverse approach to recruitment. One of the most basic examples would be: teams can’t do their own special medicals, but everyone of a certain ability undergoes a standardised test. If an MLB prospect is courting the interest of 3-4 teams, instead of doing 3-4 medicals each with different tests and camera analysis, they do 1 test, which meets all the essential basic criteria but isn’t ultra-high tech or cutting edge. The results could then be openly shared, saving every team cost but giving nobody an advantage.
Scenario 3: Persistent Expansion Model
This scenario challenges the 3 baseline statements that lead to scenario 1. Specifically, it challenges the first statement, the finiteness of knowledge. Even if knowledge is theoretically finite, the new options and ways to measure a person may be functionally endless.
So far, scouting methods have continued to fan out broadly in all sports. Data does not need to narrow scouting into who’s got the best numbers. It can broaden it if everyone is hunting for a new number that gives them a competitive edge. Early data analytics was all about points scored, and the likelihood of players giving you wins. Scouting data now can be as niche as modelling how much faster someone would be able to throw if they changed their suboptimal elbow angle they pitch with. If you work that out, you know whether it’s worth buying them and coaching that technical defect out of their game.
Thousands of “Elbow angle defect” type stats exist, some hard numbers, some soft science. To my (limited) knowledge, almost no sport has properly extracted all the available potential in psychological evaluation, game intelligence quantification or player ideology.
One real example of where this could’ve helped was Manchester City’s failed purchase of Kalvin Phillips for 42million GBP. While people have attributed part of his flop to issues with injury, he has openly stated that it took him a long time to understand Guardiola’s complicated tactical directions, and this caused their relationship to start poorly in the first weeks and months, destroying his confidence. Had they conducted robust testing to see how well players learned ideas about the game, and how detailed their thinking about tactics was… this might’ve been avoided.
The evidence for scenario 3 being the future is relatively weak, though. Beyond anecdotal evidence of what’s left to explore, you can always ask “what’s next after that”, and even the most imaginative mind will eventually run out of new ideas of ways to assess players. To play counter-counter devil’s advocate, though, people have said this across the centuries about the sciences. Max Planck’s tutor in 1878 advised him not to go into Physics because everything barring some unimportant tidbits had already been discovered. Physics, however, still had quantum to come, and still has much more to come now. If we consider player recruitment in sports to be a science, then perhaps we can be more sympathetic to Scenario 3.
Scenario 4: Competitive Niches theory
Niche theory in Ecology suggests that many species are able to survive and share resources by each having their own thing that they are optimised for. A similar idea exists in classical economics too; Ricardo’s principle of Comparative Advantage showed that even when Person A is better at everything than Person B, there will still be a thing that Person B has the comparative advantage in doing. So for example, I might be better at making watches than you, and also better at woodworking. But each hour I spend woodworking is an hour I could’ve been crafting a watch, so an hour of chair building or shelf construction costs me that lucrative watchmaking opportunity. In this scenario, you may be a weaker woodworker, but it’s still financially optimal for me to just let you do chairs, tables and shelves and buy your products off you with the proceeds of my watch sales, rather than build my own and make less money watchmaking as a consequence.
This could be true of sporting data. It is hard though to imagine a Ricardo-style situation where rebalancing is done through the sale of knowledge or end products. The closest thing I can think of to this would be the way that currently Brighton/Brentford/Bournemouth type Premier League teams develop players from abroad, absorb the risk of signing someone from abroad, then the big clubs buy the proven end-product talents at a premium. A premium that is, in essence, reimbursing the other club for their scouting/development service. Directly though, sports teams simply don’t have the incentive to sell their recruitment data. The related broader economic and ecological idea, though, is that you can survive in a competitive environment by finding your one thing you’re best at and doing it well.
In an applied sense, it may be that you decide you don’t need to try and do all the new methods everyone’s pursuing; you just need to be the best at something to access a pool of hidden talent that other people can’t assess. In baseball, this currently exists, but so does the counter-evidence (See the anecdote of baseball academies in DR in the IM Model discussion). The San Diego Padres have finished 2nd in their division to make the playoffs in back-to-back seasons, and have done so with the 2nd most conventional boots-on-ground scout-friendly setup in the league. (As per the Baseball America 2025 scout survey)
The Padres are finding success in their niche; they aren’t the only team looking at conventional intuition, with a human eye feel of a player’s skill, but they are the most invested in it. Consequently are the best at it. (Taking the Dodgers out of the equation, who cannot, on their own, sign every player they and the Padres like) This allows them to hire from their own talent pool overlooked by ‘the data guys’. They don’t need to try and have the best data analytics to compete; they can fish out of a different pool.
We are left, like with Scenario 3, to guess whether niches can exist forever, and people can carve out new niches whenever they need to, or if there is a finite number of niche potentials that will slowly reduce as we tend towards convergence.
For a niche to be stable, a club needs to be able to maintain its supremacy in that field. So for the Padres, that means they can only remain stable if the other teams can’t afford to match their traditional player analysis methods. If, say, 5 or 6 other MLB clubs all matched their boots-on-the-ground while also still outdoing the Padres at data-heavy metrics analysis, then their niche collapses.
In the event of that occurring, you have to ask: Is there a different subset of player acquisition they can disproportionately funnel their investment into to create a niche?
That question is, of course, difficult to answer.
In Conclusion
My personal gut feel, if I were forced to predict the future (which I cannot), is to fence-sit. To hedge. I think there is a theoretical possibility for a hybrid of Scenarios 1 and 3/4. I can foresee us reaching a point where new developments and ways to acquire information become incrementally less significant. A player’s eye health may be obtained at 25 and predict whether they will still be seeing the ball as well at 35… which will have a minor impact on their career value curve and predicted retirement age. But that data is the sort of thing that’s going to come down to whether a player is worth 2 million a year, or 2.1 million a year. If every team agrees on the 2 million bit, and your niche ‘only eye department in baseball’ recruitment team decides they are worth 2.1m, you aren’t going to incite a moneyball-type revolution. You probably won’t even get a best-selling book about it, let alone a film.
I do think, though, assumptions that we’ll just stagnate and never develop put you in company with the likes of Malthus, imagining we’d all die if Earth ever had 2 billion occupants. It would put you in the company of Planck’s tutor, who thought physics was finished, and it would put you slightly too close to those who decry that the end is nigh. I think the ‘end’ is very, very far off. Therefore, my personal guess is that we inch towards convergence; the gaps between teams and tactics for recruitment will narrow, unless people find spectacularly expensive methods to exclude lower-budget rivals.
I should reiterate that even if every team had the exact same recruitment data, perfectly identical, a kid still grows up wanting to play for Manchester United, not Coventry. The signing of players will be asymmetrical, both because of budgets and due to levels of appeal. Sometimes, and this may be scary if you’re into data enough to make it this far in the piece, it comes down to luck. A player’s view of themselves in high-pressure scenarios could come down to whether they got lucky on a singular play they made at age 15 trying to win a local U16 league trophy, or a summer spent with one particularly charismatic coach. Their playing value could be defined by twisting their knee walking on an uneven piece of pavement, or a star pitcher could cut their finger with a hobby drone blade.
Perhaps then, by suggesting a convergence in recruitment knowledge and insights, I’m suggesting that all the doors for plucky underdogs to beat the big guys will slowly close, or get harder to find. At least, though, if my totally unfounded gut instincts are to be believed, Physics will never be done, and the ability for a little guy to creatively out-recruit a big spender may happen less often but will never fully disappear. They’ll find that coach, that kid who grew up always wanting to play in red or the young adult whose personality shifts into a driven, hyper-focused improvement machine once given the right opportunity.