Rank: Member
Groups: Registered, Registered Users Joined: 6/14/2010(UTC) Posts: 13
|
The Equis - Momentum Indicators - Expert System in MS 11 produces good results. It appears to use a 14 bar period on MACD and RSI; however, the Stochasitc is listed only as 5-3, but no period indicated. Does anyone know if this is a 14-bar period for the Stochastic?
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers Joined: 5/25/2010(UTC) Posts: 55
|
Good results on what vehicle? What parameters did you use to do the backtest?
When I tested "Equis - Momentum Indicators" on all NDX stocks, I got poor results on each of the 98 for which errors did not occur. Out of those 98, exactly zero ended up profitable.
|
|
|
|
Rank: Member
Groups: Registered, Registered Users Joined: 6/14/2010(UTC) Posts: 13
|
I ran back test for 1 year as long-only on grop of my favorites in two ways: the formula as written that requires multiple criteria, and then I copied and edited and then isolated the components and ran agaginst the group aslo with good results. I have a few select N100's but mostly hard asset companies.
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Unverified Users Joined: 9/13/2004(UTC) Posts: 673 Location: Salt Lake City, UT
|
Stoch(5,3) - the 5 is %K periods and the 3 is %K slowing. %D (represented by the dashed line when plotting the stochastic oscillator) is a moving average of the Stochastic Oscillator Mov(Stoch(5,3),3,S) and by default uses 3 periods. In this case however, %D is not needed as the formula is just comparing the Stochastic to its overbought / oversold levels (20 & 80) and not to its "signal line" %D.
|
|
|
|
Rank: Member
Groups: Registered, Registered Users Joined: 6/14/2010(UTC) Posts: 13
|
Thanks Justin. This is helpful. I realize now in reading your response that my question is more complex than I had thought. Because the system - esp the Stochastic subset - tested well on several stocks, I wanted to test it against some systems that I was also successful using previously. Unfortunately these are only availalbe outside of Metastock (they are proprieatry systems on Thinkorswim platform - namely Persons PPS). Thus, I went to Thinkorswim to set-up something parallel to do simple comparisons to the TOS system. I thought this would be s iple comparison since Stochastics are on all charts. I was able to set (Fast, Full, or Slow) Stochastics as 5,3 but the customization also asks for period and the default period is 14 on the TOS chart and 10 on the Prophet chart. Do you have any suggestions for these settings to make something comparable? As I am sure you have experienced, it is great to find a system you like but even better when you can compare it to something you have used in the past. Unfortunately when they are on different platforms and some are proprietary, it is frustrating to be so close to a good test! Any suggestions would be helpful if you can.
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers Joined: 5/25/2010(UTC) Posts: 55
|
Alpha--one year is far too short a period of time for a valid backtest over time.
I ran the simulation again on all NDX stocks with "long" selected and out of 98 that were completed without errors, 11 were profitable.
87 were losers.
I don't see very good odds there.
|
|
|
|
Rank: Member
Groups: Registered, Registered Users Joined: 6/14/2010(UTC) Posts: 13
|
Good suggestion Mark. I was using the 1 year to narrow the field down to a few test to keep chart from having too many indicators. Now that I have narrowed the field, I can run a more robust test. So Far I like the Equis CCI (14 period; however, I got good results on 5 period too.), and I like the RSI as already mentioned as well as the Williams %R and Stoc 70/30 that came in MS11. Three questions remain before I start the bigger test:
1. Any suggestions on adequate length of time (I think it asks for periods and as you point out I put I used 250)?
2. Any thoughts on my formula question to make something comparable?
3. I found some good back testing research on using Money Flow Index but could not figure out how to make it a formula so posted that separately.
Thanks for helping me to do a better job on the testing! Much appreciated.
|
|
|
|
Rank: Advanced Member
Groups: Registered, Registered Users, Subscribers Joined: 5/25/2010(UTC) Posts: 55
|
Personally, I prefer to run my backtesting as far back as possible so as to encompass a variety of different market conditions. At any moment, market conditions may change (e.g. when VIX jumped from 15 suddenly to 22 and then up to 40+ very recently). You just don't know when this will happen and a system that worked well in the prior quiescence may suddenly be getting whipped to death now.
You can make the argument that for shorter-term trading, the market conditions stay constant but I still think you need a filter to know when they have changed. Therein lies a big challenge.
|
|
|
|
Users browsing this topic |
Guest (Hidden)
|
Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.