Intraday reversal strategy (CL) - comments welcome

Just to add a small piece, dom, have you looked at backtesting other time frames? Your pnl curve looks fitted -- this is just speculation, but from the beginning of december to today is a virtual textbook fitted curve.

If you don't get a clean, rising pnl curve before your beginning backtest period, your system is total garbage and you need to start from scratch. Even if it does work, it doesn't look very good. How many contracts have you tried?

Just the perspective of a new guy......
 
Quote from dom993:

Exactly! You have nothing to say on the topic of interest, so shut-up - there is nothing wrong for you to stay out of this thread, you have nothing to prove, and nothing for me either.
More...

see...
Quote from bone:


Also, look at how fast the 20 day historical vol and the average trading range expanded during your sampling period - it is difficult for an on-the-run range study like Pivots to cope so quickly, and your drawdowns will reflect that ( bigtime ).

My 2 cents, thanks for sharing and good luck with it. All technical studies have real limitations, so don't be terribly discouraged - and there are ways to cope.
More...

Why are you being all mean to him?
 
Quote from tobbe:

I still don't see how a computer would fail at observing the exact same things if you told it to. Unless of course, "what unfolds, how it unfolds and how it's filtered" aren't exact rules but "intuition" - but then again, you can't teach that either which was my point to begin with.

Just sayin'.
More...

Every day is different.. there is a mix of momentum players and trend followers and stops get run, pre-market news, european traders stop trading the US markets at different times every day, not to mention that the brokers are using futures for hedging their orders and only they know what their orders are.. I'm just starting to get a handle on how all the different players vector into price, and that is after a lot of screen time.. Getting it down to a computer algo is not looking like a real good way to go at this point, only for scalping, not for catching the big moves and capitalizing on them..
 
Quote from tenthousandmen:

Just to add a small piece, dom, have you looked at backtesting other time frames? Your pnl curve looks fitted -- this is just speculation, but from the beginning of december to today is a virtual textbook fitted curve.

If you don't get a clean, rising pnl curve before your beginning backtest period, your system is total garbage and you need to start from scratch. Even if it does work, it doesn't look very good. How many contracts have you tried?

Just the perspective of a new guy......
More...

I just have reliable tick data for CL, so this is all that I can test with.

The system is designed to run on volume-bar charts, I just tested V180 & V220 ... overall system performance is degraded by ~35% for V180, and ~25% for V220. The attached spreadsheet shows P&L month by month for the 3 timeframes.

I suppose this support the thesis of over-fitting - although it doesn't change much to the recent drawdown situation (complete recovery on V180, but only 55% recovery on V220)
 

Attachments

What software are you using?

Get on Ninjatrader - just try it, I think you'll find it better than most retail platforms. You can get free tick data going back to the beginning of 2011 - go to ampclearing.com and request a free demo account.

My (new guy) perspective is you are a long ways from going live and you need to learn more and try out other software too. The good news is you're definitely on the right track. My two cents is for you to keep working on this, learn about other data points, and don't stick to just one plan -- figure out ways to develop others.

If you are profitable trading discretionary (support resistance etc), then stick to that while you develop your system. If you are NOT profitable yet, do not do this. Don't do something you know you really can't......keep that bathroom mirror close by! :)
 
First of all, I'm not sure what some people here are thinking saying automation doesn't work. It is almost like being told the Earth is flat.

I would say you should simply ignore these people as they either don't know what they are talking about, or failed on the systems-side but found success (maybe) on the discretionary side. Nothing wrong with the latter but they should also know better than to say systems trading doesn't work, the former just avoid/ignore like the plague.

I tend to agree with this guy's analysis I quoted below. I think you might be on to something, but you are likely optimized and also perhaps not accounting for volatility in your risk assessment.

Perhaps a constructive path may be to test you equity curves for statistical significance?

This may be a place to start:

http://www.automated-trading-system.com/bootstrap-test/

Quote from sheepsucker:

Hi,

I have actually seen a similar thing in a strategy of mine, not the same instrument but in a correlated one.

Some general comments:
It appears based on your comments and the results, there could be an element of curve-fitting present.
However, there could be an edge, BUT maybe it is not as big or reliable as the backtests indicate.

Hope this helps a bit!
More...
 
Quote from bwolinsky:

The 1-tick slippage should be applied in 2 tick increments to all backtesting. It is a mistake to only include 1 tick slippage when they are blanket values to use so that you don't get overtraded results.

Use convert to market orders, and also add 2 ticks of slippage to it.

I have no way of deciphering drawdown from these values, pf for target 1 is obviously a lesser and more frequent target than the others. The system P/F and P/F for target 2 appear to include overnight trades.

Limit order execution guarantees against any validity in this backtest.
More...

This newbie doesn't get my review of profit factors was as much about drawdown as the less than 1:1 win loss ratio I'm sure this system has.
 
My thoughts:

1) No setups work 100% of the time, so you need to identify win% of the setup based on stop and target selection.

2) Once you have developed that you have to apply standard statistics. For example, lets assume a system has a 50% chance of winning or losing. Can you have a string of winners or losers in a row even say 10 of them, of course that is possible. For example, you could play 10 hands of black jack using basic strategy and win all hands. So you need to use math to determine if your draw down falls outside of the realms of probability or not.

3) If your draw down is within normal parameters, then there is nothing wrong, and you move forward. If your draw down is not within normal parameters, then it points to the fact you have a bad setup and you need to go back to step one which is finding an edge. An edge being defined as a trade setup that is profitable.
 
sounds like 123/ross hook. Pretty sure you can buy an off-the-shelf coded system for it for any broker, its that prevalent
 
''Meanwhile, I was frantically scrutinizing the charts, looking for clues. But really, I couldn't find any "reason" for that loss of performance, aside from using "bad" pivots. ''

The clue you`ve missed is,that a pivot serves as a support/resistance line as well as a BO line.It will always be that way.You can backtest until you puke,but that is.
 
You must log in or register to reply here.
Back
Top