Matrix Games Forums

Have a question for the Puresim veterans. I am having a great time with this product. Hats off to Shaun & co., for what is truly a quality product and a must-have for any baseball fan.

I am replaying the 1974 season with the Cardinals. I am managing each Cardinal game and simming the non-Cardinal games. That year, Lou Brock set the record for steals in a season with 118. While I have played 31 games into the season he has stolen 12 and been caught 6. He has not been getting on base a great deal so the opportunities to steal have been few.

My question is..his speed rating is 71 in the game for 1974. Is this the only rating that is used in evaluating ones ability to steal bases in Puresim? I have run into 10-15 other players in the Natinal League alone that have much higher speed ratings. Guys like Concepcion, D. Baker, etc that never stole the bases Brock did. Hell, Craig Robinson a journeyman ss for the Braves has a 100 speed rating.

While I normally do not fiddle with editing players, etc. I am wondering if this speed rating should be higher. I would be interested in anybody's thoughts on this. If you replay seasons and you notice something that may be "out of wack" do you take the liberty of changing the ratings?

Thanks.

Jliven61

It seems to me that the program assigns ratings for historical players within a "range" or applies some sort of criteria I don't understand.

In any event, I have fiddled with ratings when I think they are seriously out of whack. While waiting for the final version of the patch, I have, for example, boosted the "arm" ratings of all catchers in the league in order to cut down on the number of successful steals and steal attempts. It has worked pretty well, although it's only an interim solution until Shaun gives us the "real deal" (which I hope works as well). He mentioned that he is changing the criteria by which speed is rated, which worries me a little, because I don't think that's the problem (and won't that have an effect on attempts to tag up and advance on fly balls?).

I had 1956 Mickey Mantle rated as an average center fielder (!) with a range of 35. Believe me, that got changed in a hurry. On the other hand, Ernie Banks had ratings that qualified him as an excellent shortstop (which, despite his great offensive talents, he never was). I toned him down a smidgen defensively. Nellie Fox with an "eye" rated in the low 40s? Forget about it. Babe Ruth with a power rating of 68? Nonsense. 1968 Bob Gibson with stuff 59, velocity 64, control 48 (!), endurance 58? Gimme a break.

I would have goosed Lou Brock's speed to 100 immediately before even starting league play (sounds like maybe his hitting ability is underrated, as well, unless you think the Cubs were right ...). 71 is ridiculously low for him (particularly considering that the game doesn't seem to take into consideration a player's savvy, quickness, and natural deftness).

When I start an association using historical players, I always review everyone's ratings to see if there are any that are really ridiculous (and there always are).

The patch should improve accuracy in many regards. That said, its only a refinement. It is extrememly tough to build an algorithm that looks at E, PO, AS, Age, Position, SB etc. And do a "perfect" job of assigning subjective ratings. There is just not enough source data.

That said, I have been heads-down coding for 2 weeks now and 1.1 beta 2 will have a bunch of brand new heuristics it is considering in both modeling player ratings and on-field performance.

It has been a very tedious 2 week period but I think I am emrging with an even better historical simulation, and PureSim Gold 1.0 was pretty darn good.

The toughest part is sitting and waiting for 25 year simulation stretches just to see the impact of a minute tweak to the engine (which is gettin closer and closer to DIPS by the way).

Anyway, beta 2 will get posted this week so start a season with it and tell me what you think.

Have you set his steal frequency to steal more frequently in player usage settings? You can also set manager tendencies to steal more and turn other players down to steal less or never steal. He should get lots of attempts then.

ORIGINAL: puresimmer
That said, I have been heads-down coding for 2 weeks now

Sounds like my honeymoon oh, so many years ago.

Thanks for your work, Shaun. I really appreciate your dedication to improving this game (which doesn't need much in the way of improvement, as far as I'm concerned). The ability to edit players gives me the flexibility to "correct" ratings, so I don't think that improvement in the program's assignment of those ratings is a big deal.

After this round of patching settles down, though, I plan to hit you up with some ideas about how the game's graphics presentation might be improved ...

Have you set his steal frequency to steal more frequently in player usage settings? You can also set manager tendencies to steal more and turn other players down to steal less or never steal. He should get lots of attempts then.

That won't have an effect, as he manages the games personally.

Jliven61, I'd say its your game, do what you want with it - if that means 'bump his speed up', then do it!

And yes, I think 'Speed' is the only factor used for the player's stolen bases, its pitted against the 'arm' of the catcher and the 'holds runner' of the pitcher.

Yeah Amaroq I think I will. I was just going to hold off if the steal success took into account other factors other than those you listed.

Thanks.

Jliven61

You're welcome! Have fun - and now I'm curious. Let us know how many he bags for you, will ya? [;)]

No problem. I increased his speed from 71 to 92. So far 12 outta 18 in 32 games, thats way off the pace. But thats okay too. I'll keep ya posted.

Jliven61

OK, I looked into this. Lou is a bit of an outlier...

One of the factors PureSim uses when attempting to assign the subjective SPEED rating to real players is their age. Lou was actually 35 when he had 118 steals (most likely one of the oldest steal leaders ever!)

Also, the algorithm does not just look at the one season, it checks out how the player did as far as steals from a few previous and a a few future seasons as well.

I'll look at tweaking the algorithm to weight steals a little more...

Alternately, you could special case exceptionally high single-season values of steals to ignore those factors. For example, it looks like anything beyond 70 steals in the modern era is pretty impressive, and only 7 people have broken 100 since 1900, rising to 13 at or above 90, 22 at or above 80, and 42 at or above 70. (vs 58 70+ steal players between 1887 and 1899).

In the modern age, we have four players 31 or over with more than 70 steals:
Lou Brock, 35, 118 - never stole more than 56 after that
Ron LeFlore, 32, 97 - never stole more than 36 after that
Maury Wills, 32, 94 - never stole more than 52 after that
Ron LeFlore, 31, 78 - got better the next year!

A 'special case' that dropped all of those fellows into something like
Min{100, Max[(90 + ((SB-70) / 4), previousAlgorithm()]} would get you what players will expect for speedsters in the modern era.

You could also 'special case' aging for speedsters as well, with exceptionally high speed (80-100) dropping to (65-80) after age 32, so you'd get something like Lou Brock 35 = 100, Lou Brock 36 = 80, Lou Brock 37 = 65, which seems to work with his two subsequent seasons (both 56-steal seasons - still impressive, but not dominant.) If it continued to drop at about that rate, you'd have Lou Brock 38 = 50, which is again about right for a 35-steal season (especially given how frequently he was caught that year), and Lou Brock 39=40, which might correlate to a 17-steal season.