[DEV] ISFL Forums
*Value Oriented Breakkdown on the CHI/PHL Trade - Printable Version

+- [DEV] ISFL Forums (http://dev.sim-football.com/forums)
+-- Forum: Community (http://dev.sim-football.com/forums/forumdisplay.php?fid=5)
+--- Forum: Media (http://dev.sim-football.com/forums/forumdisplay.php?fid=37)
+---- Forum: Graded Articles (http://dev.sim-football.com/forums/forumdisplay.php?fid=38)
+---- Thread: *Value Oriented Breakkdown on the CHI/PHL Trade (/showthread.php?tid=20195)

Pages: 1 2

*Value Oriented Breakkdown on the CHI/PHL Trade - iStegosauruz - 03-17-2020

[div align=\\\"center\\\"]Background and Methodology – [/div]

Today we had one of the first trades of the season. Chicago shipped their first and second round draft picks in the upcoming draft to Philadelphia for Farley Hank (RB 750TPE), Johnson Harding (DT 308TPE), and Philadelphia’s third round pick in the upcoming draft. In summary the trade looked like:

Quote:“PHI receives:

S22 CHI 1st round pick
S22 CHI 2nd round pick


CHI receives:

Farley Hank RB
Johnson Harding DT
S22 PHI 3rd round Pick


PHI retains Hank and Harding S21 contracts”

And can be found here.

The trade sparked a significant amount of controversy, especially as it related to the value that Chicago received in the trade. This is a win-now move for the Butchers. They’re committing to trying to force open a window for contention, a decision made riskier by the upcoming draft being one of the deepest in history.

I decided to try to explore the value of the trade for Chicago, and tangentially for Philadelphia. In doing so, however, I had to establish benchmarks for the value of draft picks. This is going to be a topic I plan to explore more in depth in hopes of creating a draft value chart similar to the various ones used in the NFL, including the Jimmy Johnson Chart, Chase Stuart's AV Chart, or the Harvard Value Chart.

Creating a value chart similar to those requires a LOT of effort. Most of them require calculating the “approximate value” for players over their career, which is something I’ve done in a smaller sample size when I looked at the approximate value players generated in Season 20. You can find that here if you haven’t read it yet. Doing that for a multiple season sample to create a draft value chart is going to be a massive undertaking and I decided that for brevity and to get this post out as soon as possible so the trade was still fresh that I needed to take a different approach.

What I ended up doing was throwing together a rough value chart that looks at the amount of TPE a player generates. I then looked at the rate at which teams “hit” on a player who generated peak TPE of a decided benchmark. The benchmarks I chose were 250TPE – which I deemed as playable, 500TPE – which I deemed as above average, and 750TPE – which I deemed as excellent.

I chose to pull the data from the Season 13 through 16 drafts. This is notable because Season 15 and Season 18 are known for being particularly deep drafts because of the influx of reddit players. I chose to not include the Season 18 draft because I wanted players to have had at least five years of TPE earning to factor in. It is also worth noting that I think in theory player retention and TPE earning has probably increased as seasons have progressed, so your chances of finding good players in later rounds has increased since the seasons I chose. In my estimation I think this is probably due to league development and understanding on how to approach recruitment, but all of this is just theory. In short what you need to know is that these are baseline numbers that I can’t imagine can be lower but I can see a definite world in which they’re higher. If they’re higher the draft picks the Butchers gave up are more valuable, meaning they need to get more value from the players they received back in this trade.

Also of importance before diving into the numbers is that when I looked at the value Chicago received I chose to measure it in win percentage. I traded the two players they acquired to them in the sim engine and then set their depth chart based off what I formulated was the strongest combination of players they could put into various positions. This left Farley Hank starting in two formations – the Spread and Shotgun – where the Butchers kick their primary RB into the flex. He was the backup in the I-Form, Pro, and Two-TE sets. I considered moving him into various TE or flex positions but decided I’d go with what I thought was an average estimation of his usage.

I plugged Johnson Harding into the lineup in every depth chart on defense. In some situations, he replaced Trae Bacon (DT 159TPE), while in others he replaced a bot that Chicago is using on their defensive line.

When simulating the games, I chose not to simulate any matchups that pitched Chicago against Philadelphia. The variable simulations where Chicago is utilizing their new players leave Philadelphia with a particularly wonky depth chart that requires more changes to optimize. I figured that for analysis sake I could overlook matchups between these two teams. That also meant I did not simulate any matchups between them in the control study – where Chicago has their original lineup and Philadelphia still has the players they traded away. I figured that was the best way to consistent and comparable statistics.

I chose to have an approximately 6400 game sample for Chicago with their new additions and to have the same sample for Chicago with their original depth chart. This broke down into 200 games where Chicago was the home team and 200 games where Chicago was the away team against the other times in the league – since I excluded Philadelphia. This means there are 1600 games at home for Chicago with their new players and 1600 away. The same breakdown applies to the control study where they have their original depth chart.

[div align=\\\"center\\\"]Draft Analysis – [/div]

Across the four drafts I chose to survey the peak TPE value of the average first round draft pick was 774.47. The average second round draft pick had an average peak TPE value of 640.33 while the average third round pick had an average peak TPE value of 379.33. This can be seen expressed in this chart:

[div align=\\\"center\\\"][Image: JWdr29p.png][/div]

You can do a fair bit of faulty analysis with this data. On face you can say that Chicago sent picks that had a combined potential peak TPE value of 1414.8 to Philadelphia in return for a package of players with 1059TPE and a pick with potential peak TPE value of 379.33. This is a total of 1438.33TPE. In this line of analysis Chicago gains 23.53TPE in value, however that isn’t a great measurement.

It is important to note that Hank Farley (RB 750TPE) will be entering his first season of regression after this season. Assuming he gains between 0TPE and 100TPE for the remainder of the season he would be left with between 600TPE and 680TPE for next season. This is one of the factors why it is hard to use the above measure of TPE value exchanged to evaluate the trade.

What can make it is easier is looking at the amount of times a team “hits” on their pick in various rounds. Recall how I subdivided peak TPE reached by the draft picks into 250TPE, 500TPE, and 750TPE tiers. You can break down the number of trackable players – meaning I was able to find a measure of peak TPE for them – in the first three rounds of each draft and compare that to the number of players who met each benchmark. That analysis represented in a series of charts looks like this:

[div align=\\\"center\\\"][Image: ll5tP05.png]

[Image: Hg8yrV5.png]

[Image: kbZSM3j.png][/div]

Chicago is giving up two chances to potentially get players with strong TPE peaks. Four of the nine trackable players in the Season 16 first round, for example, had peaks over 750TPE. Six of those nine players had peaks over 500TPE. The reddit draft included in these samples – Season 15 – is in theory a good representation of the value that Chicago gave up in these two picks. In that draft a whopping 80% of the players drafted in the first round – eight in ten – have reached a historical peak TPE of at least 750. Of those eight six are over 1000TPE while one is only 13TPE away from that benchmark. In reddit classes first round picks net high quality players.

Just to provide more data as well I dug deeper and examined the hit rate of players across the entirety of the draft – not just the first three rounds. That should better represent deep drafts – which the upcoming one is expected to be – and is represented in this chart:

[div align=\\\"center\\\"][Image: R9yWbAw.png][/div]

I think the best way to look at the value of the picks is to look specifically at the Season 15 draft, however, to highlight the potential low risk end of the transaction for Chicago I calculated some probabilities based off the average hit rate for the various benchmarks across the four drafts in the sample.

A pick in the draft has an average chance of 31% to pull a player with a peak of 500TPE. When you map the probability that either one of the picks OR both of the picks pulls a peak 500TPE player the equation looks like this:

[div align=\\\"center\\\"][Image: 0y2jCMV.png][/div]

What this means is that the probability they pull a peak 500TPE in a scenario like this is 52.39%. You can map the same probability with a 750TPE player:

[div align=\\\"center\\\"][Image: gMttgCC.png][/div]

This means that Chicago had a 36% chance of pulling a 750TPE player with one or both of their picks. If you choose to look at it through only the lens of Season 15, Chicago had a 64% chance of pulling peak 750TPE player with both their picks and a 96% chance of pulling at least one 750TPE from one of the picks. Recall that in Season 15 six of the ten picks in the first round have reached a peak TPE of 1000. In the second round eight of the ten picks have reached a peak TPE of 1000. This puts the probability that Chicago drafted two peak 1000TPE players at 48% and the probability they get at least one at 92%. All of this assuming this upcoming draft resembles that particular draft.

That is a lot of value that Chicago is giving up for one 750TPE player about to hit regression and one 300TPE player – essentially the value of a third round pick.

[div align=\\\"center\\\"]Game Value Analysis -
[/div]
To find the value Chicago received in this transaction I chose to look at the change in win percentage that occurred for them after executing the trade. As I explained in the methodology this resulted in an approximately 3200 game sample for them without the new players and an equivalent approximately 3200 game sample for them with the new players.

It is important to note that as always there is some minor error in sampling due to the occasional program popup on the laptop I use for simulation that results a few more sims or a few less sims. I run the sim engine on an almost eight year old laptop that has seen better days.

When I modeled the control situation – which was Chicago before they acquired the new players – they won 49.22% of their home games and 24.94% of their away games. The chart of results:

[div align=\\\"center\\\"][Image: oGVB1cS.png][/div]

When I modeled the variable situation – which is Chicago’s depth chart after they acquired the players from Philadelphia – they won 50.31% of their games at home and 25.56% of their away games.

[div align=\\\"center\\\"][Image: oGVB1cS.png][/div]

This breaks down into Chicago having a winning percentage of 37.08% across 3200 games without their new players and a 37.95% winning percentage across 3200 games with their new players. This is an improvement of 0.87%.

[div align=\\\"center\\\"][Image: IDr4PsE.png][/div]

[div align=\\\"center\\\"]Conclusions -[/div]

1. Chicago gains a minimal increase to their expected winning percentage – excluding games against Philadelphia - on the year for two draft picks in a loaded upcoming draft.
2. Those draft picks have a traditional 52.39% chance of pulling at least one peak 500TPE player and a 36% chance of pulling at least one peak 750TPE player.
3. In a more loaded draft – as this upcoming one is expected to be – those picks have a 64% chance of pulling two peak 750TPE players and a 96% chance at least one does. They have a 48% chance of pulling two peak 1000TPE players and a 92% chance at least one does.
4. Although I lean towards thinking this is a bad trade for Chicago it is not my place to pass judgment – I only wanted to crunch the numbers on the value they gave up and the change in winning percentage they expect to gain. I would love to see their internal numbers from before making this trade and comparing them to mine.

[div align=\\\"center\\\"]Notes – [/div]

1. The winning percentage they expect to have with their new players does have a small margin for error depending on how they move players around the depth chart. Based on some experimental moving I did in my spare time it doesn’t have a huge effect either way depending on how they move those players.
2. If any GM is looking for similar work done on a trade they’re considering I’m always happy to crunch numbers – shoot me a message on Discord.
3. I’ve heard through the grapevine that Chicago is going to make more moves. This interests me on two fronts. First – I’m curious to know what they use as capital to make more moves. In a loaded draft their 3rd, 4th, and 5th round picks are of more value than they would be traditionally, however, so were the 1st and 2nd round picks they just gave up. If they gave up their two highest value assets to have a minimal increase in winning percentage I can’t imagine they make a move that creates a substantial gap or opens their window into contention considerably more than it already is.
4. As always I make all my data open source. You can check it in this folder.
5. The turn around on getting this out same day was crazy. Got it in right under the wire.

*Value Oriented Breakkdown on the CHI/PHL Trade - Frostbite - 03-17-2020

This deserves so much money

*Value Oriented Breakkdown on the CHI/PHL Trade - gucci - 03-17-2020

Word heavy. Nice.

Another great analysis, keep killing it.

*Value Oriented Breakkdown on the CHI/PHL Trade - Jiggly_333 - 03-17-2020

[Image: image0.jpg]

*Value Oriented Breakkdown on the CHI/PHL Trade - Kyle - 03-17-2020

Can we just give this man all the money already?

*Value Oriented Breakkdown on the CHI/PHL Trade - Troen - 03-17-2020

(03-16-2020, 09:42 PM)Kyle Wrote:Can we just give this man all the money already?

Agreed!

This post makes me genuinely sad I can't pay money directly to other players because it's amazing.

*Value Oriented Breakkdown on the CHI/PHL Trade - Isidore94 - 03-17-2020

Lot of numbers here. As a guy who writes 90% of media on numbers, I approve of this post.

*Value Oriented Breakkdown on the CHI/PHL Trade - br0_0ker - 03-17-2020

Great read first off.

Secondly, does the testing you did factor in the position change part that this trade was made in mind with? And did you run any offset probability for the 3rd round pick we acquired, for CHI's expected chance to pull a 750-1000 TPE player?

*Value Oriented Breakkdown on the CHI/PHL Trade - iStegosauruz - 03-17-2020

(03-17-2020, 12:17 AM)steelsound Wrote:Great read first off.

Secondly, does the testing you did factor in the position change part that this trade was made in mind with? And did you run any offset probability for the 3rd round pick we acquired, for CHI's expected chance to pull a 750-1000 TPE player?

Thanks for the feedback!

As I explained in methodology I inserted Farley Hank into y'alls depth chart as a starter in two formations and a backup in the other three. If you tell me how you intend to swap players around I can definitely run the simulation again and get you the numbers - or you can provide me your internal testing numbers although I'd still likely back check them. Generally, if you'd like to contend differently its helpful for me to see
As I also stated in the piece I did experiment with different variations - swapping Leaf to WR and TE in various spots/moving Farley around where it seemed like the advantageous thing to do. I didn't save any of the numbers because they were smaller sample sizes and it was more for margin of error testing, but it didn't have a substantially greater impact. I'd still, on whole, say you gave up immense value.

As for the third round pick. With the four season sample I used a third round pick has a 28% chance of pulling a peak 250TPE player, a 23% chance of pulling a peak 500TPE player, and an 18% chance of pulling a peak 750TPE player. If you narrow the scope to only the Season 15 draft its a 50% chance to pull a 1000TPE player and a 70% chance to pull a 750TPE player.

I also went ahead and pulled data for the Season 18 draft's third round. The average TPE of a player in that round is currently 413.6. You have a 10% chance of pulling a 750TPE player and a 0% chance of a peak 1000TPE player if you narrow the sample to only that draft. They haven't had as long to accrue TPE and some are on pace - for example one has 634TPE - but on whole four of the ten picks in the round are still sub 300 and another is sub 400. It was not nearly as strong of a reddit draft.

Basically - I can see some logic for pulling the trigger on a risky trade but it doesn't seem like in any world you got appropriate value. Giving up draft picks in a strong draft for a player who at best going forward past this season will be 680TPE, a 300TPE DT, and a third round pick which is still good in this draft but not nearly as good as a first or second isn't much to get and doesn't seem as viable a strategy as building traditionally and having two blue chip players at the top.

Your top four TPE players are already regressing or will be after this season. I understand the idea that the window is now but your expected winning percentage before the trade was 37.08%. Its going to take a lot of big moves to turn that into an Ultimus contender - and the point is I don't think this trade was that type of move even with any potential margins for error in my calculations, especially considering the capital you gave up. Hence the article - based on a wholistic analysis of the situation you probably should have kept the draft picks and pivoted for the future.

*Value Oriented Breakkdown on the CHI/PHL Trade - Mooty99 - 03-17-2020

iStegosauruz going 1st oa, throwing that out there now!