Rosie2167 wrote:paul8210 wrote:Nah, there already are enough clues in mystery leagues.

What clues do you use for figuring out pitcher cards as we might be missing something.

There are indeed lots of clues. The trick is to

(1) decide what the key differences are between the cards that you want to analyze? Where are the biggest discrepancies in various hits or outs? (In order for you to actually use a difference between cards, it has to be a significant enough difference that it could be distinguishable, one card from the other, in the card results. A difference of 10% in die roll occurences in the cards is not going to be distinguishable in the actual mystery game results. Find the big differences between the different cards, by left or by right handed batters. Also, if the pitcher is a lefty, you may need to analyze a card difference versus RHB, since you might not have enough game results versus lefty batters to be a large enough sample size.

(2) Make sure that the game results aren't unduly skewed to batter card readings, especially early in the season. If they are skewed to batter card readings, you'll need to wait until your sample size of game results is bigger. Give the poor guy some more innings and die rolls!

(3) After you've selected the key card result variables (#1 above), is it something that strat already tracks? In my experience, strikeouts are often a good differentiating factor for pitchers, and is of course great, because strat tracks it! (See my original post on this thread).

(4) If the key card result variable isn't tracked by strat, get to work accumulating the data on actual game occurrences versus the correct handed batter. Add them up. (This is only fun if you are obsessive-compulsive. Strat could be used as a test for this condition, perhaps

).

(5) Calculate the game occurrences per 216 batters faced - i.e. take the game occurrences and multiply by 216, and divide by the number of actual plate appearances by that hand of batter against your pitcher. Strat does track that on the lefty-righty stat page.

(6) Assess (I do this judgementaly, not with precise analysis) which of the five mystery cards is most likely represented by these game occurrences.

You could use homeruns, or ballpark homeruns, as a clue. The problem with these clues is that they do not typically have a large number of die roll occurrences on the cards. You can use this sometimes, especially when it is (possibly) a card that gives up a lot of these. The problem with this approach is - especially with a small sample size - that it could be related to batter card results. This is especially true for situations that favor the batter - i.e. a lefty pitcher giving up lots of homeruns to RHB, or a righty pitcher giving up lots of homeruns to a lefty batter.

Walks are somewhat better, but the problem with walks is sometimes intentional walks (I always set my IBB to conservative in the mystery games), but also the same problem as with homeruns. With a small sample size, you have to be aware that your pitcher might have been facing batters who draw lots of walks. (Of course, in the mystery game, you don't know what batter card he is facing, esp. early in the season).

Of the results that strat tracks, I find strikeouts can often be the most useful way to differentiate cards. It is often the case that a pitcher will have more strikeouts in the earlier years of his career. But you need to look at the 5 actual mystery cards to see which years have more strikeout results, by counting the die roll chances. Strikeouts versus the opposite handed batter (i.e. lefty hitter results on the card for a RHP card) are often very informative, as they can sometimes vary very significantly from one card year to the next. But any discrepancy or variation between card years is helpful, if the differences are significant enough. I like to use differences of 10 die roll chances or more to evaluate pitcher card differences. Again, if the sample size is small, you need to have a correspondingly larger card and game result difference to make a reasonably confident conclusion from that.

If you can't do it based on strikeouts, you might be able to use other game results, such as popouts, foulouts, lineouts, etc. The problem with flyballs and groundballs is that they are so numerous, and the co-mingling of Xchart results in the game results data. With popouts you also have X chart results from Catch card X outs co-mingling, as well as some BP single out results. With lineouts, you have BP single out results co-mingling. And of course you always have the batter card results mixed in, with every kind of analysis. BUT, you can notice a measurable difference between cards based on game results with popouts and lineouts - again if the differences between the different mystery cards for that pitcher are significant enough in the first place. So for example, in the Pedro Borbon 1970s mystery cards, his popout results versus right handed batters are as follows, on each of his 5 mystery cards:

Borbon vs RHB:

1974 - 4 popouts (i.e. die roll chances with this card reading)

1975 - 5 popouts

1977 - 9 popouts

1978 - 16 popouts

1979 - zero popouts

With a large enough sample size, not skewed to the batter card, you should be able to add up the popouts by RHB, gross them up to 216 (216/PA by RHB), and ask yourself, "Is this number explainable based on each of the various cards, within a reasonable tolerance?". To answer this question, you need to consider possible batter card results (on average maybe 2 per 216), catchers card X out results (maybe 2 per 216, on the high side), and possible outs from BP single chances (maybe 1 popout per 216), for a total of maybe 5 popout results per 216 from other sources. 4 popout results from other sources is probably closer to the average, but say 4 or 5. If your total popout occurences per 216 in the game results is 25, you can safely assume that you don't have 1974, 1975 or 1979. And there's a high probability that you have card 1978. If, on the other hand, the game results are 10 per 216, it's probably not possible to make a conclusive evaluation. If the game results per 216 is 5 or less, you can conclude that you don't have his 1978 card, but maybe not much else for sure.

If you combine this analysis with some others, (e.g. for Borbon, doubles and triples by each of lefty and righty batters, per 216 would be a good way to differentiate cards), and look at the statistics as a total package, you might be able to narrow it down further. Don't forget about singles as a way of differentiating cards. Because it's so prevalent, though, you need to have a big card difference. Once you have doubles and triples accumulated, then you have the singles, by calculation. And don't forget about the state of your fielding, especially the outfield, when considering the occurrences of singles and doubles. Be wary of making card conclusions based on triples, unless the discrepancy is striking, because of the possibilities of X chart results or runners taking extra bases, even on a base hit (i.e. stretching a double into a triple).

I know this sounds overwhelming - but if you isolate a couple of card differences for a particular player, and focus on those differences, it is possible to do this.

And now you know why I would like strat to at least keep track of doubles and triples yielded on the L-R stat page!