Free Agency Strategy Question

[GoldenCrest Games]

Interesting thing I've been wrestling with during free agency AI development:

In the NFL, why don't every team chase the same free agents? It seems simple at first, but it's hard to reduce to something measurable. In Desktop Dynasties, I've rated every player on the amount they improve a roster, the amount of improvement per dollar of salary, and the overall value of a player per dollar. It seems reasonable that these would be useful in deciding how to spend your FA dollars, right?

What I'm finding is that by using these measures, almost every team should be bidding for the same group of players. After all, really good players will improve every team. Good players with reasonable salary demands would be a bargain to everyone. Functional cheap players would also be a good bargain for teams that have to fill roster spots.

However, in the NFL they don't all chase the same players. I at first chalked it up to the value calculation, but that doesn't differ wildly between teams. It seems that most teams in the NFL have close enough player value calculations that it's not a major difference, and doesn't significantly throw the calculation off, so it has to be something else.

I'm getting a little too far in the weeds, so I need to pull back and gather some other insight.

Any ideas?

[QuikSand]

Really interesting weeds to get into, I think.

One quick thought: a computer game affords us a much clearer picture of a player's "value" then we would be likely to have in real life. Let's think about CB Prince Amukamara, who is heading to free agency this offseason. 28 year old, former 1st round pick, an NFL starter with a fair body of work.

In FOF, I could see his ratings (give or take a little scout error) in everything that matters to me - with no effort. In the real world, that's a lot harder, isn't it? Maybe we can grab stats like interceptions, catches allowed, passes defensed, tackles, etc -- but what story does that tell for us? Can he handle press coverage? What if the receiver is tall? What if he matched up against a 245-lb tight end? What if he's asked to switch coverages into the slot? What is he has to read the offense and pull into press after the offense tip off its play design - does he read/react appropriately? Can he blitz? Can he read out of a called blitz? There could be 250 more questions appended to this list... right?

Anyway...I think the bottom line is that NFL teams consider this a matter of limited resources. Most more or less figure they have enough cap space to address a certain (fairly short) number of needs in the early going with higher-profile players, and do their homework there. Maybe 3-8 NFL teams will consider "starting-caliber CB" to be a priority for big money free agency, and will look at a guy like him in detail to start answering those questions. But not 32. So, at some point we will hear that the list of teams expressing in PA seems like it's these thre or four teams, then he gets an offer from one, and he either signs it or it lingers a little while to allow other teams to get into the mix. That seems to be how it works.

Shorter answer... your discussion seems to suggest a lot more precision than I think exists in a real game of so many unknowns and hard-to-measures.

[QuikSand]

The former economics teacher in me also would say that to the extent "the market" clearly understands what a player's value is... then teams should be more or less willing to bid right up to the point where they are paying exactly the right amount for his services. There shouldn't be any "deals" in free agency... a guy who is good enough to contribute 0.4 wins should get paid accordingly.

Uncertainty is what makes that assumption laughable. Guys lose motivation after signing a fat contract, they get hurt, they get old, they rebound from injuries, they find a better system, they leave their perfect system, and untold numbers of other such stories that make sports free agency frought with tons of boom/bust stories.

What do you think DE/LB Junior Galette would be worth on the open market this year? (I didn't look up whether he is, but he strikes me as a perfect candidate for widely-varying valuations by different teams who have views on character issues and injury risks)

[GoldenCrest Games]

Originally Posted by QuikSand

Shorter answer... your discussion seems to suggest a lot more precision than I think exists in a real game of so many unknowns and hard-to-measures.

I think you're on to something here. I don't model unknowns or "risk" in any meaningful way yet. Every player is perfectly scouted and each team's needs are perfectly understood.

Maybe that's the missing factor?

[JPhillips]

You also have to take into account the players already on the team. The free agent that adds minimal value compared to your current roster is worth less to that team than an objectively less valuable player that adds more to that roster spot.

[bhlloy]

Yeah I think there are a few things here, most of which Quik touches on

1) Scouting variance - the concept that in video games we know what a player is rated in clear attributes that are easily correlated to how they perform (plus or minus a scouting fog of war usually) whereas IRL there are stats and measurables that have a much wider range of interpretation from team to team
2) Human error - some GMs and front offices just suck. Hard to model in a video game but definitely a huge factor
3) Stylistic preferences - some teams really value some players because they fit or have a perceived fit in a particular scheme. Some teams are much more willing to take a chance on younger players or players who have been backups with perceived potential, whereas some teams are much more likely to look for proven vets.

As usual, FM is way ahead of the market and tries to implement all of those into the game, with varying levels of success.

[Carman Bulldog]

Originally Posted by JPhillips

You also have to take into account the players already on the team. The free agent that adds minimal value compared to your current roster is worth less to that team than an objectively less valuable player that adds more to that roster spot.

I think this is one of the bigger factors. Let's say your starting LT is a 76 while there is an 84 rated free agent LT. Your starting RG is a 65 and a free agent RG who is rated a 75. Which one do you target?

Obviously, the better player is the 84 LT and that is also, generally speaking, a greater position of value. At the same time, it could be argued that the 75 RG could potentially improve your team more.

One also needs to consider schemes. In this instance, with a right-handed quarterback on a pass heavy team, that LT becomes more important. However, if the team is more run heavy (and particularly inside runs), then the RG has greater value.

Now let's say the offense actually runs a zone run-blocking scheme and there is also a 73 rated RG free agent. He lacks the size and power of the 76 RG but has good feet and is athletic.

Of the three free agents, the 84 LT, the 75 RG and the 73 RG, the lowest rated player arguably becomes the best fit.

[AlexB]

Also need to build in the cap situation: how much money is tied up in a particular position, and how much dead money cutting the incumbent player creates, will help determine which players a team will go after

[GoldenCrest Games]

Good stuff so far guys!

I'm getting closer, but it's still not quite right yet. Here's where I am right now.
Teams evaluate each player based on:

1. Total improvement to the roster. This is measured against the current depth chart, and where the player best fits in.
2. How much money the player wants, compared to what the coach would pay for that type of player. This is heavily influenced by play-style and roster-style for each coach.
3. How many roster spots the team needs to fill. If the team has a lot of open spots, the coach won't spend as much for one player.
4. Improvement-per-dollar threshold. This is a limit that any player needs to meet for an offer to be made. Coaches won't overspend for little improvement.
5. Value-per-dollar ratio. This is a way to make sure the player is actually worth the salary. (prevents bidding wars from going too high)

All of those are used, but it's really the total improvement that sorts the targets. Everything else is used to filter the list.

The contract offers look ok now, it's just the actual targeting of players that needs work. The "what would I pay" (#2) rating has started to help spread out offers, but it's still more bunched than I'd like.

I don't want to make it too complicated, but I think there's something I'm missing.

[GoldenCrest Games]

Quick update:

I threw in some indicators to actually see how teams are valuing players in real-time, and it looks about like we'd expect. (even as players start getting signed, and rosters fill up)

The biggest issue right now is budgeting. I think a few of you were going there with the comments, and it's looking like a major factor now. What I've been seeing is that teams are all following the same process:

1. Find your favorite player, and bid on him.
2. Take the remaining money you can spend (based on roster space, cap space, dead money...etc) and find your favorite player that fits within that amount. Bid on him.
3. Repeat #3, but with whatever is left over after that guy.
4. Repeat until you have no money or roster spots left, or can't find anyone you like for the price.

The problem with this strategy is that you're already really top-heavy in offers, and there are very few middle-half players getting bid on. Every team wants to hit that homerun, and blows the FA money in step 1 and 2. (usually $70-$90 contracts) After that, there are only $1-$5 contracts left for steps 3,4. Those $20 - $30 guys never get bid on.

Budgeting seems to be the answer. Design a salary strategy that buckets contracts into tiers, and force the team to manage each tier separately in FA. Right now, the numbers look like the star player tier would be 65% of the cap, and would be 1 to 5 roster slots. The good player tier would be 35% of the cap and would hold 2-6 roster slots. The rest of the roster slots (4-10 slots) would be filled with the remaining 5% of the cap.

Coaches would differ on how many slots they reserve for each tier, which would drive how much they will spend on a single player in that tier.

Seems like it's a step towards simulating how teams bucket money between stars, average, and filler players, right??

Any thoughts?

[JonInMiddleGA]

Originally Posted by GoldenCrest Games

The problem with this strategy is that you're already really top-heavy in offers, and there are very few middle-half players getting bid on. Every team wants to hit that homerun, and blows the FA money in step 1 and 2. (usually $70-$90 contracts) After that, there are only $1-$5 contracts left for steps 3,4. Those $20 - $30 guys never get bid on.

Wait though ... doesn't this get dealt with to some extent in the lather/rinse/repeat process that should take place after each big ticket player comes off the board?

Those mid-tier guys aren't going to get a lot of offers early on (or if they do, they aren't max realistic dollar offers), that's kinda realistic to me. (edit to add) But their offers should begin once money gets freed up for all but one of the teams who bid on jumbo salary guy X.

Another element to this -- maybe? -- is that their earliest offers are most likely coming from teams who have fewer roster spots to fill with FAs that are above replacement value. They don't have as much money tied up in bidding on as many "$1 - $5 guys", so they should have enough left to make initial offers of like $7 to those $20 guys that aren't getting offers early.

[lungs]

Originally Posted by Carman Bulldog

Obviously, the better player is the 84 LT and that is also, generally speaking, a greater position of value. At the same time, it could be argued that the 75 RG could potentially improve your team more.

One other thing to consider... Where does the 76 LT grade out as a RG? Sign the 84 LT, bump the 76 LT over to RG and get two birds stoned at once.

[Vince, Pt. II]

It also seems like perhaps the algorithm should figure out a budget strategy for the team before bidding? I mean, if I'm a team with say 8 major holes to fill, I'm not going to blow my entire budget on one top-tier FA. Then again, some might think that's the best way, I guess.

[GoldenCrest Games]

Originally Posted by JonInMiddleGA

Wait though ... doesn't this get dealt with to some extent in the lather/rinse/repeat process that should take place after each big ticket player comes off the board?

Those mid-tier guys aren't going to get a lot of offers early on (or if they do, they aren't max realistic dollar offers), that's kinda realistic to me. (edit to add) But their offers should begin once money gets freed up for all but one of the teams who bid on jumbo salary guy X.

This makes a lot of sense, so I took a look. Since many of the teams are trying to get the homerun, they are often in bidding wars for a small handful of players. Players wait to sign until the bidding wars are over, but that's causing them to stay on the market for a long time. That is causing large amounts of money to get tied up from many teams, and not enough teams are moving on to the next tier of players.

It looks like teams that bail out on the star players are going after mid-tier players like I'd expect them to. I have to get teams to bail out quicker, or escalate the bidding war quicker, to get everyone to move on.

That's a good one!

[GoldenCrest Games]

Originally Posted by Vince, Pt. II

It also seems like perhaps the algorithm should figure out a budget strategy for the team before bidding? I mean, if I'm a team with say 8 major holes to fill, I'm not going to blow my entire budget on one top-tier FA. Then again, some might think that's the best way, I guess.

I think this is also a problem I've been having. Teams need to be more proactive in setting budgets and knowing what they can spend, before making bids for players.

This could also help create differentiation between roster styles and coaches. I could see one coach who likes to spread the salary around while another goes for the superstar, no matter the cost.