Each July, Fangraphs releases our annual trade value series, highlighting the top 50 players in baseball and considering contract status and performance. I’ve been in charge of this exercise for the past four years and have some connections with the rest of the Fanggraphs staff and team. This year, I have company. Fangraphs readers will create their own list of trade values for the first time.
I am proud to announce the debut of our crowdsourcing trade value tool. Created by Keaton Arneson and developed by Keaton and Sean Dolinar, the tool briefly summarizes the question of “Which of these two players do you like?” to create a composite ranking. Using this tool is easy. When you bring it up, you will have two players appear and ask you to choose the transaction value you think has a higher value:
Ah, but what does “higher value” mean? Sometimes the simplest questions are the toughest questions. Having higher trade value is not the same as a contract with younger, younger, and more team contracts. Of course, this could be some combination of these things, and other factors. In the real world, players vary in the level of value to a team based on many considerations, from holes left by recently injured stars to where the team finds their place in the playoffs. Promising prospects may mean more to rebuilding the club, just as a reliable disparate person might move the needle for a team with October ambitions. We can’t tell you how to weigh these factors, and it’s part of the fun of building a list of trade values in the first place. What we can do is provide some data we think is in definite and let you decide how to apply it.
To do this, we provide biographies and statistics for the players you choose. For each player you are asked to make a decision, you will see their current age, stats for the 2022-24 season, stats for the 2025 season, and contract details. For players who have not made their debut before 2025, we also provide board data: Prospect rankings, ETAs, personal tool ratings and their future value ratings.
In each case, your instructions are simple: just select the player you think has more trade value. It means it’s all up to you; we’ve said a lot about how players value each other in the trading deadline. There is no formula for selection, no mathematical certainty. If anything, this exercise won’t be particularly fun. This is more important than science, but as we found in contract crowdsourcing, we suspect there is some wisdom here. Select the winner and you will get another match. You can perform tasks multiple times (or very few) as you wish. (I will also notice here that you can’t reset or deselect.) The more choices you make, the more our system will understand the preferences of Fangraphs readers. Yes, it does mean you need to log in to use the tool. This is how we work alone.
Two things you notice when you make more choices. First, we provide your personal top 10 list. This is based only on your own choices and ignores the rest of the readers. Second, you will notice that the game is not completely randomly selected. You’ll often see players you’ve never chosen before, we think the showdown between players you’ve rated and the need for more in the database. We found that the 150-200 showdown is a good cutoff for creating a reasonable individual top ten, but you should do a lot of it at will.
Behind the scenes, we will convert your A-OR-B selection to ranking using a modified form of the ELO rating system. Elo’s core is to predict the work of the winner of the showdown based on the current ratings of the two competitors. After the game, ELO adjusts the score based on what actually happens. The winner gets points, while the loser loses the same amount. The point is that the number of points exchanged depends entirely on the surprising results. The more results you have, the more points you will get. In other words, the harder the opponent is, the more points you get to beat them, and the more points you lose. The opposite is true: the easier the opposition, the fewer the victory will be scored.
If the Dodgers win a game between the two squads, Elo will not change its estimates for the Dodgers and Rockies, taking the team, not the player, as an example. But if the Rockies get rid of their frustration, Elo will improve its rating. How much bigger? This change is proportional to the possibility of depression. In the expected 80%/20% distribution duel, the popular winner (80% score) has only a small reward. However, if the loser wins (20% result), they will receive four times the reward.
One or the error of Elo’s differentiation depends on the function you are using – is the path depends on the path. The earlier showdown provides odds for future showdowns, which means the order of the game is important. Consider a simple example: You choose twice between Gunnar Henderson and Bobby Witt Jr., who ranked first and second in last year’s trade value list, selecting each. If you first choose Henderson as the winner without previous information, the engine treats it as an even showdown and gives him some points. Then, when you choose Witt the second time, Elo thinks it’s a frustration – after all, Henderson ranks higher than Witt due to your preferred choice. Elo ruled to score more than he lost in the first round, while Henderson headed the dock with the same amount. Now, Witt is ahead of Henderson in your rankings, but if you choose Witt to win the first game and Henderson wins the second game, the reverse will be right.
This may be useful for evaluating the strength of the team – we care more about the games that happened yesterday than the ones that happened in April or even 2024 – but the concept of “early” doesn’t make much sense for this exercise. To remove this path dependency, we will completely randomly select the order of matches and calculate the ELO score based on this. We then repeated the process 100 times, selecting a random order at a time, which gives us 100 different sets of ratings. We then average scores produce paths that are impossible final order.
Worrying that someone will fill up the ballot box? do not be afraid. Once all matches are submitted, we will check to make sure we are not getting spam and weighting if necessary to balance feedback from those who submitted a lot of matches instead of just submitting some matches. In this way, we believe we will measure the perspectives of Fanggraphs audiences well.
So, please help us! Jump to the tool and start picking. The tool is now live and can be used in the conclusions of next Friday and the Trade Value Series. After that, we will post a summary list of the crowd and dive into some of the most interesting games. Pleasant Clicks – Remember, whatever you think the “higher value” means, use it as your standard.
