Odds Czar: Simple Biases in the Futures Markets 2005

Great Thread Art!!

I programmed your bias rules and tested them on 10 years of SPY (Buy/Sell next day at open with commission based only on bias rules, no protective stops). I know your bias rules are for futures, but I trade stocks so I tried them on SPY.

Your basic system was profitable. I also optimized it using (-1, 0, or +1) as the weight for each of the bias rules. Got some interesting results.

Here’s a table of results for the baseline bias rules and some of the optimized systems. Some of the systems had better performance with SPY if some of the rules were ignored (weight = 0) or reversed (weight = -1):

 

I guess it wouldn't let me attach the gif file. I'll just paste the results in text format:

System: Baseline
Rule 1: 1
Rule 2: 1
Rule 3: 1
Rule 4: 1
Rule 5: 1
Rule 6: 1
Rule 7: 1
Rule 8: 1
Total Net Profit $14,842.42
Max Intraday Drawdown ($15,630.97)
Percent Profitable 61.35%
Profit Factor 1.08
Return on Account 94.96%
RINA Index 28.75

System: B
Rule 1: 1
Rule 2: 0
Rule 3: -1
Rule 4: 0
Rule 5: 1
Rule 6: 0
Rule 7: -1
Rule 8: 0
Total Net Profit $66,720.85
Max Intraday Drawdown ($13,531.51)
Percent Profitable 50.00%
Profit Factor 2.24
Return on Account 493.08%
RINA Index 48.11

System: C
Rule 1: 0
Rule 2: 0
Rule 3: -1
Rule 4: 1
Rule 5: 1
Rule 6: 0
Rule 7: -1
Rule 8: 0
Total Net Profit $75,388.18
Max Intraday Drawdown ($9,853.62)
Percent Profitable 46.93%
Profit Factor 1.84
Return on Account 765.08%
RINA Index 40.57

System: D
Rule 1: 1
Rule 2: 0
Rule 3: -1
Rule 4: 0
Rule 5: 0
Rule 6: 1
Rule 7: 0
Rule 8: 1
Total Net Profit $78,424.47
Max Intraday Drawdown ($13,313.47)
Percent Profitable 64.27%
Profit Factor 1.48
Return on Account 589.06%
RINA Index 121.93

 

Art
Could you clarify rule #3 . The rules refer to the highest close of past 50 days and the table of biases refers to the highest high.

Also where is the exit point?

Rules
3. The highest close of the past 50 days occurred before the lowest.

Indicator
50-day highest high

thanks bnun

 

I also appreciate Art showing us his trading stuff, but I'm not convinced you proved anything, other than perhaps proving that his rules don't work as well as buy and hold? I'm not trying to knock your work, just trying to get a handle on how it, and Art's rules, can be helpful.

You showed on your first system, that using the rules with a straight bullish-bias did worse than buy/hold over the last 10 years.

And on the others, you optimized by discarding some rules and actually reversing some others to get a decent return. I just don't see how any kind of stable system can be developed here.

What am I missing?

Regards.

C

 

i v trader wrote

in theory you could accomplish your first question, although i don't know why you'd need to jump the gun like that. as for the second question, certainly that too is theoretically possible, and i've tinkered around with that in research (like i've looked at most everything else under the sun). these particular signals don't lend themselves to any intraday bias though.

 

setharb wrote

by the time i wrote my first book, i was trading several successful systems. my big struggle was i was also trying to trade on a non-system basis at the same time. (a whole myriad of psychological reasons behind that). that endeavor was a disastrous failure, culminating in an actual psychological screening in which i was pronounced a compulsive gambler and told to quit my profession.
(VERY big crisis point in my life).


my best personal success is interday systems that tend to flip signals every few days--i'd say about four on the average. i've also had success with holding a position over one night, and if i were long, selling at previous day's high or buying in on the low for short. (old larry williams trick). i have some daily stuff that works--the challenge there is overcoming trading costs with your relative small trading arcs.'
everything i create in a primary market of interest (like the s&ps) i want to see work, at least to some degree , in other markets. so i will include markets like currencies and bonds even though, at the moment, i'm not personally trading them. their signals are also valid.

 

brun12 wrote

oops. it's mean to be the highest close verses the lowest close.
again, these are just basic building blocks here--no integrated system at this point. so all the biases mentioned suggest entering on an open, exiting on the close.

 

"And on the others, you optimized by discarding some rules and actually reversing some others to get a decent return. I just don't see how any kind of stable system can be developed here.

What am I missing?"
------------------------------------

Inquisitiveness; desire to explore, test, and experiment with new ideas.

I see nothing wrong with testing new ideas and rearranging rules to see what happens. I learned a few things about the market doing this data mining exercise, so it was very beneficial to me. And that’s all that counts.

I didn’t optimize; I used the optimization engine to test the discrimination potential of different combinations of rules. The MA and lookback window parameters remained the same. I’m not developing a system here; from my perspective there’s a big difference between data mining and system development.

Again, I thank Art for sowing some great ideas.

 

No problem. Thanks, Art, for your thoughts.

And thanks Wolf for doing the analysis. But you did present the different combinations as having "profits" of X%, with specific drawdowns, etc, like they could be viable systems. Sorry if I got that wrong.

All I can see is that you re-arranged randomness. I guess, as an automated system user and builder, I'm always searching for ideas that make money. Maybe with more work, walk-forward testing, montecarlo, etc, there is something there. JMHO, but I'm skeptical.

Again, please don't get me wrong, I'm not knocking you I'm just trying to glean some useful info that I can incorporate into my own designs. Isn't that what we all are here for?

Good luck in your system efforts and good trading to all.

C

 

Hi Art,

I am not a trading lecturer or book author, just a no-body trader who has been trading the ES everyday for the last few years.

I agree with most of your 8. observations, but disagree with a couple of your remarks.

1. I think it is absurd to say the average trader can't succeed unless he is 100% mechanical. I think if you watch the same market day in day out, every tick, over years you can get a strong sense or market feel for the movements of that market. Although, it is good to trade by pre-defined rules based on extensive research and testing sometimes you have to think outside the box. I personally think the best approach, IMHO, is a discretionary trader who generally follows pre-defined rules, although you may view this as the worst of both worlds?

2. I don't think a one size fits all approach is necessarily the best. You state a trustworthy idea must test over a number of different markets. I hold the view that it is better to specialise in trading one market, become an expert in that market and develop a niche for yourself.

These are just my humble opinions. Feel free to disagree. One last comment. I feel that daytrading is the hardest type of trading and the reason for that in my view is the shorter your trade timeframe the higher the intensity of emotions you feel in terms of fear and greed. Again, just my own experience/observations.