You may want to read Robert Pardo's excellent book "The Evaluation and Optimization of Trading Strategies". He gives a very good explanation of how to statistically determine the answer to your question.

I think you need to define your terms first? What constitutes failure and what is success? If you break even in a down market is this success? Do you have a standard benchmark to measure against, maybe the whole of the SP? Or maybe you have a basket of successful stocks already? Thanks, David

Often developers will suggest that so long as the draw downs do not breach the max draw down in the back test than the model is intact. I am unable to embrace that concept completely as I prefer to look forward in addition to only the past.

Understanding the trading strategy by knowing it's sharpe ratio, drawdown from backtesting and volatility of assets during the period of backtesting will help to form stop-loss thresholds. If they are breached, it is prudent to exit. If the strategy begins to trigger stop-loss exits frequently, there would be reason to investigate if it has stopped working.

Everyone is worried about the strategy and the outcome. I been seeing everyone post about negative strategy and the best way to analyse, . Does not having to worry about any of that, a strategy. EA that does all that you are asking, is what I have. isn't that what an EA is supposed to do?

Apparently a Robot is the way to go, it wins more than it looses, why is that, not a strategy?

for a strategy to be a failure then, it must have been taught of a few months ago. or a few years ago.

In front of you , you can own 1 license of a, Robot that has been in development for over 12 years.

we are,a software company., not a Financial Service. Nicolas Gomez

drawdown, standard deviation, Sharpe ratio and it's length of time should match in back test, your strategy logics should perform in a same length of time as it performed in simulated results. if you see huge differences then logics are not competitive to meet all complexed behaviors of market in certain timing

Many strategies work a couple years and then start to deteriorate. You need to catch that moment. For getting some measure about how far a live strategy is allowed to deviate from the backtest, I've published a formula here:

http://www.financial-hacker.com/the-cold-blood-index/

When your bank card is suddenly swallowed by the cash machine!! As simple as that!

I hope you have taken transaction and impact costs in your back testing. there are many strategies which give excellent sharpe ratios in backtesting but when put in real time, the impact costs (and unaccounted for bid/offer or sharp event based market moves) eat away all profit. If that's whats happening then immediately exit the strategy. Else persist with it until strategy s/l is hit (Always better to keep a hard stop for the strategy as a whole)

Trade cost, spread, slippage is an issue even professionals cannot handle and are heedless, it is an invisible silent poison which destroys the profiting face of strategy. its solution is medium term trading to keep most minimum no. of trades (only highly filtered refined trades of strategy where historical probability suggests higher win level

You still need to know and be intimate about how the strategy works. How it is supposed to produce the alpha. If it is a 'black box' system, you wouldn't know if it is failed until after the fact (after the losses have already occurred). This is sample selection bias at its "best".

Scott Boulette, does the book layout complete systems with teardowns as to how you would perform the backtest and example pitfalls/workarounds?

We've written a few articles about this topic. The main focus is knowing your trading statistics and detecting changes in those statistics. We use rolling statistics for this for faster detection. Here are a couple of guest posts on this topic: http://www.seputarforex.com/abe/products/view.php?id=266843&title=trading_strategy_failure_the_biggest_threat_to_your_trading_account and http://www.seputarforex.com/abe/products/view.php?id=263651&title=how_to_succeed_with_a_high_win_rate_trading_system

Mathematically. The obvious start is to work out what is your expected mean return and the expectedvolatility of those expected returns. I don't think many traders ever do this, but it is an objective way to assess whether a drawdown is most likely just random market movements or a failure of the model.

For a different take on this subject, you should know that I'm publishing in Q1, 2017 a two volume series, "The Strategic Analysis of Financial Markets," that deals with this subject in Volume 2, Chapter 16. Basically, I offer advice and quantitative methods for dealing with this problem. Check out "The Tax Day Trade," and the mathematical, "A Lottery Money Pump," for examples.

Regularly compare your performance against peers or indices reflecting similar strategies. If your strategy starts failing you should see consistent underperformance vs your comparisons.