What are the Best Stocks to RTM Trade?

In2Deep · Feb 16, 2011

I don't understand how you can trade an RTM strategy on anything other than an intra day time scale--trading the volatility off the open.

But I do understand how you can really get screwed trading this way because positions move against you very fast and then move in your favor just as fast.

empee · Feb 16, 2011

1) High Volatility

2) Lots of public participation (GOOG, AAPL, etc); lots of amateurs trading small lots versus big institutional names trading it.

intradaybill · Feb 16, 2011

Quote from dwpeters:

There is more than one way to trade reversion to the mean. What I am doing doesn't require margin and it hasn't broken down in the past 4 years of live trading or 10 years of backtests. Then again it is not scalable to the fund level either.

Sounds good. How long do positions last on the average?

dwpeters · Feb 16, 2011

Quote from intradaybill:

Sounds good. How long do positions last on the average?

About 3 days. Larry Connors has done some good work in this area. I don't use his services or tools, (and would never pay thousands for his seminars) but I recommend his books.

intradaybill · Feb 16, 2011

Quote from dwpeters:

About 3 days. Larry Connors has done some good work in this area. I don't use his services or tools, (and would never pay thousands for his seminars) but I recommend his books.

It sounds like you are very experienced in this area. Recently I had an idea I don't know whether it makes sense but I am working on it. I use a program to find patterns (Price Action Lab). The normal use of the program is to set a high enough profit factor and have it find high probability setups. I was thinking that since some patterns continue to work in the future but some do not and basically their win rate mean reverts if I could find a way of identifying those patterns I would then fade them. In this way, I can have the two extremes: High probability for momentum and low probability for mean reversion. Do you see any value in such startegy? Till now I stayed away form RTM strategies because I didn't know how to properly set stops but with patterns I will know (because the system does it for me).

dwpeters · Feb 16, 2011

Quote from intradaybill:

It sounds like you are very experienced in this area. Recently I had an idea I don't know whether it makes sense but I am working on it. I use a program to find patterns (Price Action Lab). The normal use of the program is to set a high enough profit factor and have it find high probability setups. I was thinking that since some patterns continue to work in the future but some do not and basically their win rate mean reverts if I could find a way of identifying those patterns I would then fade them. In this way, I can have the two extremes: High probability for momentum and low probability for mean reversion. Do you see any value in such startegy? Till now I stayed away form RTM strategies because I didn't know how to properly set stops but with patterns I will know (because the system does it for me).

I am not clear on what you are proposing. If you think the momentum pattern is low probability then you would fade it? That doesn't sound like RTM, but whatever can give you an edge is worth exploring. I view RTM like a stretched rubber band. The farther price is pulled in one direction in the short term, the more forceful it tends to snap back.

I am a big believer in diversification of strategies. My backtests and live trading show a huge diversification benefit from combining uncorrelated or negatively correlated strategies. This is the advantage of end of day trading. Since I generally don't manage my positions intraday I can take dozens of positions across multiple strategies.

And regarding stops. If a momentum stock turns down it likely will no longer look like a candidate for a momentum trade, and so it makes sense to close the position. If an RTM stock pulls back further, it still looks like an RTM candidate, so where to get out? I find no stop level that improves results. Doesn't mean there isn't one, but my trading/testing is not sophisticated enough to find it. I manage risk with small position size and diversification.

goodgoing · Feb 18, 2011

Quote from intradaybill:

It sounds like you are very experienced in this area. Recently I had an idea I don't know whether it makes sense but I am working on it. I use a program to find patterns (Price Action Lab). The normal use of the program is to set a high enough profit factor and have it find high probability setups. I was thinking that since some patterns continue to work in the future but some do not and basically their win rate mean reverts if I could find a way of identifying those patterns I would then fade them. In this way, I can have the two extremes: High probability for momentum and low probability for mean reversion. Do you see any value in such startegy? Till now I stayed away form RTM strategies because I didn't know how to properly set stops but with patterns I will know (because the system does it for me).

I already do this.

Break search sample in two parts: Past and current. For example past is 1997 - 2001 and current 2002 - present.

Set in Price Action Lab the win rate > 95% and trades > 1 and run the search in current sample

Select patterns with win rate > 97% and trades < 18

Test in the past sample. Select those patterns that fail in that sample.

Those are the patterns that are most likely to be hit by a mean reverting win rate towards 50%.

I tell you, if you start trading this way, you will never go back to other methodologies. It takes a lot of work but it is almost the perfect edge. Opposite to what most people are doing. By the way, this will not work with indicators, only with price action. It works well with intraday patterns too.

GoldStandard · Feb 18, 2011

Thanks for sharing this, GG! Mighty generous of you.

If I understand you correctly, you're finding patterns with a worse than 1 to 1 TP/SL ratio, so that when you fade them and their win rate reverts towards 50% you have a strategy with a near 50% win rate but a favorable risk/reward ratio?

Quote from goodgoing:

I already do this.

Break search sample in two parts: Past and current. For example past is 1997 - 2001 and current 2002 - present.

Set in Price Action Lab the win rate > 95% and trades > 1 and run the search in current sample

Select patterns with win rate > 97% and trades < 18

Test in the past sample. Select those patterns that fail in that sample.

Those are the patterns that are most likely to be hit by a mean reverting win rate towards 50%.

I tell you, if you start trading this way, you will never go back to other methodologies. It takes a lot of work but it is almost the perfect edge. Opposite to what most people are doing. By the way, this will not work with indicators, only with price action. It works well with intraday patterns too.

intradaybill · Feb 19, 2011

Quote from goodgoing:

Select patterns with win rate > 97% and trades < 18

Test in the past sample. Select those patterns that fail in that sample.

Those are the patterns that are most likely to be hit by a mean reverting win rate towards 50%.

Finally a good idea was posted here other than fighting over nonsense political and economic issues.

But, with a maximum of 17 trades, the maximum possible win rate, other than 100%, is 16/17 = 94.11%, cannot be greater than 97%. Thus, no patterns will meet this criterion.

Are you then looking for 100% win rate in that period?

I am sure if I can find some I will immediatelly fade them. That 100% cannot be sustained for long.

Edit: Gee, thanks. I just figured it out. You just increased the possibilities of Price Action Lab by a factor of 10 or even more. Thanks.

GoldStandard · Feb 20, 2011

Why less than 18 trades? Is it just because more trades than that would mean the pattern is likely to hold?