Statistics of a good system? | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Statistics combined with system. Profitable? What do you think 71 replies

Grid system based on optimization and statistics 1 reply

Alternative Reward to Risk Ratio for Trading System Statistics 16 replies

The statistics on your best viable system? 0 replies

Trader Statistics and Broker Statistics 0 replies

Statistics of a good system?

  • #1
  • First Post: Jan 4, 2011 3:07am Jan 4, 2011 3:07am
I'm currently teaching myself to program and in the process backtesting several very simple strategies.

I understand basic data issues, unreliable backtests, curvefitting, sample size etc.

My question is what would the statistics be of an acceptable mechanical system i.e. if we accept a backtest as 100% reliable, what would maximum acceptable drawdown and minimum expected payoff levels be.

E.g. Would a system with a payoff of only 0.6pips per trade really be tradeable in the actual market?
  • #2
  • Jan 4, 2011 5:29am Jan 4, 2011 5:29am
As long as you have a system with a positive expectancy, primary, this is considered a good start.

The system returns vary proportionally with the max. drawdown you might expect.

A good trend following system with a positive expectancy can be expected to have a max. drawdown ranging between 30%-50%. Such a system is likely to generate consistent 30%-40% in annual compounded return. Of course it differs from a system to another.


Regards,

Nader
  • #3
  • Jan 4, 2011 8:57am Jan 4, 2011 8:57am
Quoting pasopvirpot
My question is what would the statistics be of an acceptable mechanical system i.e. if we accept a backtest as 100% reliable,

Any back testing can be only 90% reliable.


Quote
what would maximum acceptable drawdown and minimum expected payoff levels be.


I would say any drawdown over 50% will be maximum for most strategies, but can be exceptions.


Payoff per month it's up to you. You can start with 5% and then Sky is not a limit but Risk is a subject for decision.
  • #4
  • Jan 4, 2011 11:59am Jan 4, 2011 11:59am
Quoting pasopvirpot
I'm currently teaching myself to program and in the process backtesting several very simple strategies.

I understand basic data issues, unreliable backtests, curvefitting, sample size etc.

My question is what would the statistics be of an acceptable mechanical system i.e. if we accept a backtest as 100% reliable, what would maximum acceptable drawdown and minimum expected payoff levels be.

E.g. Would a system with a payoff of only 0.6pips per trade really be tradeable in the actual market?
0.6 pips a trade is not enough. Just take a bit of slippage and maybe some spread increase, then your winning strategy will turn into a losing strategy. You gotta have a comfortable cushion in order to be able to make a strategy actually viable in the real market.
  • #5
  • Jan 5, 2011 1:50am Jan 5, 2011 1:50am
Quoting Custos
0.6 pips a trade is not enough. Just take a bit of slippage and maybe some spread increase, then your winning strategy will turn into a losing strategy. You gotta have a comfortable cushion in order to be able to make a strategy actually viable in the real market.
Thanks for all the replies. For a scalping system what would be the minimum pips per trade to aim for to be able to overcome slippage and spread changes?
  • #6
  • Last Post: Jan 11, 2011 2:11pm Jan 11, 2011 2:11pm
Quoting pasopvirpot
Thanks for all the replies. For a scalping system what would be the minimum pips per trade to aim for to be able to overcome slippage and spread changes?

That is not a simple question. The general guideline is not to risk more than 3% of the available account equity per position. You can pick any amount below that and be safe. The target profit has to be at least equal to the amount risked in my opinion. That is a Risk/Reward ratio of 1:1. But remember the reward portion has to take the spread and slippage into account. You can control that to some extent by your choice of trading times.
0 traders viewing now