[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Optimal F



PureBytes Links

Trading Reference Links

A couple things I've never quite understood about Optimal F:

1./ The formula for Optimal F that I've seen is determined by the
sequence of wins & losses expressed in points gained/ lost per trade.
Wouldn't it make more sense for the formula to use percentage gains/
losses instead?

2./ I'm not quite sure how to apply Optimal F on the equity side.
Assuming that your number of contracts to trade is 10 (= Equity/ f$,
where f$ = MaxLoss/ Opt F), does this mean that you should buy 10
shares? Or does the current price of the stock come into play somehow?

BTW: most folks object to Optimal F because of the drawdowns intrinsic
to good systems. This is not my major beef with it, though.

The real issue as far as I'm concerned is that of how stable a system's
optimal f values will remain in the future. 

To test this I downloaded a track history of an advisory service off the
Internet (400 trades +) & started calculating optimal f's on a rolling
30 trade basis. I got as far as 50 periods before I got bored.

Here's what I found: optimal f's ranged from a low of 0 to a high of
0.78201. Here are the numbers if you want to chart them yourself:

0.00000
0.03757
0.14097
0.19020
0.07821
0.22591
0.11447
0.05552
0.00000
0.01749
0.00000
0.00000
0.00000
0.00000
0.00000
0.00000
0.00000
0.00000
0.00000
0.00000
0.00000
0.02658
0.00000
0.00000
0.00000
0.03115
0.20641
0.22222
0.26666
0.39491
0.43672
0.33340
0.26971
0.23848
0.17571
0.18767
0.20966
0.39248
0.51253
0.54132
0.60125
0.72341
0.70673
0.77279
0.78201
0.63866
0.61184
0.60271
0.64078
0.55246

Here's the equity curve if the optimal f for the next trade is based on
the previous 30 trades:

 $38,537.50 
 $39,725.00 
 $42,650.00 
 $35,825.00 
 $40,375.00 
 $29,975.00 
 $24,462.50 
 $21,962.50 
 $21,962.50 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,800.00 
 $21,625.00 
 $21,625.00 
 $21,625.00 
 $21,625.00 
 $23,275.00 
 $33,337.50 
 $39,937.50 
 $55,137.50 
 $65,287.50 
 $46,312.50 
 $41,712.50 
 $40,825.00 
 $9,825.00 
 $17,025.00 
 $13,450.00 
 $23,762.50 
 $27,012.50 
 $49,262.50 
 $62,200.00 
 $84,950.00 
 $82,537.50 
 $157,337.50 
 $243,587.50 
 $38,887.50 
 $34,237.50 
 $38,550.00 
 $60,487.50 
 $16,987.50 
 $7,787.50 

Well, if I were really gung-ho I would go beyond my fifty trials & maybe
even repeat the process w/ different sample sizes .... I did do a simple
trial, however, where I just split the trade history in two: optimal f
was 0.51 for the first 215 trades & 0.73 for the last 216 trades.

In short, optimal f numbers do not seem to be very stable over time.
This dramatically increases the likelihood of trading at levels beyond
the point of inflection, wherein lies disaster. 

The only solution I can see is to increase the size of the trade samples
until optimal f levels stabilize. (It's only my assumption that they
will stabilize.) Or perhaps include a process for trading the inevitably
wild equity curve.

OK. Don't forget the questions that started this email :-) ...

Thanks,

Cab Vinton
cvinton@xxxxxxxxxxx