just found this here (that post had no "recommendations" before I gave it one ...).
Ending my Acorda pick now may ultimately have a negative impact on my CAPS score, due to the artifactual inflation of points in outperforms that were initiated at low prices. Intuitively this seems appropriate, but in fact it does not reflect the actual market. For example, if I now rate Acorda as underperform, pick it up as an outperform again if it drops to 28, and then ride it back up I will gain points in the short term but eventually become relatively negative once the price hits about 40. That's because my outperform points only reflect my most recent start price, not the original one. That's how I lost my score leader position in Progenix despite having three strong picks so far. However, I think it's better that I approach CAPS with my actual investment strategies rather than as an attempt to inflate a meaningless score which is only a weak reflection of the positive predictive value of a pick.
I have felt somewhat guilty of doing this "score point inflation thing" for quite a while. While the "caps" game "accuracy" is rather ridiculous the score point part of the "ranking system" is not all that great either. Mainly for not being "annualised" at least to some extent. This makes "catching up" very hard and what is worse, keeps players from changing their calls.
An illustration. The stock price of a certain "caps" game "ratable" stock rises by 200% in a certain time period of length T, then drops by 50% in the following time period (of size T again) , then goes through this "cycle" "over and over again".
Player 1 makes an "outperform" call on the stock at the beginning of the first period and then "reverses" the call at the beginning of the following periods.
Player 2 makes an "outperform" call on the stock at the beginning of the first period and that's it.
Let's say the benchmark is constant (and again assuming no dividends that could make things more complicated).
For T=1 and p(0) = 1 you get this development of the scores p1(t) and p2(t) of players 1 and 2.
Now let's see what would happen if players 1 and 2 invests the famous $100. Let's ignore "commissions", taxes, ability to buy and sell any fraction of a stock, fees & interest for the "short sales" and all that.
Player 2 simply puts $100 into a long position in the stock and holds. Player 1 has two separate accounts, the "trading account" and the "cash account" puts $50 into a long position in the stock (in the trading account) and puts $50 into the cash account cash. At the beginning of period 2 (t = 1) he sells ($150, stock is "up" by 200%), puts 50% of the $100 gain into the cash account (so now he has $100 in that cash account) and "puts the remaining $100 into a short position" (i.e. sells short shares, receiving $100) in the stock. He covers that short position at the beginning of period 3 (t = 2) and again adds 50% of the gain (i.e. $25) to his cash account (now at $125) and puts the rest ($125) into a short position and so on. Maybe that was not as clear as it could have been. So player 1 is always 50% invested. In the odd/even periods he is invested in a long/short position in the stock. His separate cash account after each "50% of gains" cash transfer has other half of the money, the not invested part. So he is actually "playing" this pretty conservatively.
This shows the sum of the "account balances" after the "cash transfers" (both accounts have that balance at the time, so player 1 has twice that amount (all in cash) before making the next trade) for player 1 and the trading account balance for player 2.
(log is the natural logarithm.)
So as to the "caps" game score points player 2 (buy and hold) clearly "outperforms" player 1 "after a few cycles". As to "investing success" the situation is "different" ...