Who Here Backtests Manual Strategies ?

[moonmist]

I personally think that there's so much *context* used with discretionary strategies that back-testing would be a wash. It's not really back testable if you cannot codify it - and if you can, it's probably not discretionary anymore.

In science, the results of one lab should be reproducible in other labs. It is also true for automated or program trading.

I wonder if this is true for discretionary trading. By the way, what is the definition of discretionary trading ?

Just my two cents

 

[wrbtrader]

In science, the results of one lab should be reproducible in other labs. It is also true for automated or program trading.

I wonder if this is true for discretionary trading. By the way, what is the definition of discretionary trading ?

Just my two cents

There are two types of discretionary traders...

1) Traders that are not automated but they use rules to make their trade decisions (some of these rules can not be coded).

2) Traders that are not automated and use no rules. Thus, they rely heavily upon their trade experience and/or intuition.

Two traders using the exact same discretionary trading method will not be able to reproduce the exact same trading results for obvious reasons involving crazy number of variables that's trading related and personality related. Nature just makes us different...individuals.

Simply, we will just react differently with the same trade method especially whenever markets change...a change that occurs often every month. There's a lot of trade method threads here at ET that proves such is fact involving a shared trade method and a lot of psychological tests/books that you can read in which explains how "people" react differently when put in the exact same life situation or professional situation.

Heck, even folks using the exact same trade method (discretionary or automated) will produce different physical reactions in their own body...all of which has influence on the next trade or next trading day.

 

[Tavurth]

In science, the results of one lab should be reproducible in other labs. It is also true for automated or program trading

"The cotton industry engages experienced buyers who purchase cotton. Their procedure is to pull a tuft of cotton out of a random bale of a lot. They will look at that tuft and feel it, tease it out, listen to the crackling perhaps as they do so, touch it with their tongue, and through this procedure they will determine the class of cotton the bales represent. There are about a dozen such classes. As a result of their decisions, purchases are made at certain prices; blends are made in certain proportions. -- Now these buyers cannot yet be replaced by the Machine.” -- Asimov

I think experience allows recreation of similar ability, although it may not be based on currently quantifiable variables.

 

[i am nobody]

There are two types of discretionary traders...

1) Traders that are not automated but they use rules to make their trade decisions (some of these rules can not be coded).

In contrast, manual testing implies your visually reviewing each historical trading day without any automation/codes to do it for you.

Exactly! This applies to me.

 

[qxr1011]

I personally think that there's so much *context* used with discretionary strategies that back-testing would be a wash. It's not really back testable if you cannot codify it - and if you can, it's probably not discretionary anymore.

agree

more to that: trader tests not only his trading rules and methods , he is testing himself

in a real life, while watching situations developing, there are many urges to do something what should not be done (or because of loss of concentration or hesitations, the moment when something should be done is missed), that's when trader's guts, balls, and brains are tested...

you can not replicate that in back-testing

trading is like sex, you can not back-test yourself

 

[globalarbtrader]

From the gate I traded live with real money. When testing strategies I would just trade with very small positions.

But going forward I want to test out strategies and develop various edges.

Paper trading is an option but I don't want to wait 3-4 days to see if the trade would have played out in my favour.

Are there any backtesting software programs or online that you guys would recommend for a discretionary trader.

I don't need anything fancy just something that is literally place trade, click play and watch 3 days of price moves in one second.

Preferable mac but I have access to windows also.

This is for FX only

I think you have to stop using the term "back test" for discretionary trading if you want to avoid confusion. In most peoples language a back test is for systems only. In a back test you take the past, and say what would system X have done in a given situation. You then see how things work out. Obviously this is repeatable over many different situations, and when you repeat the process on the same data you would get the same result.

What you're talking about is essentially "practice trading", but not in real time. You get the past and say what would I have done in that situation. You then see how things worked out.

Congrats on digging out of your drawdown BTW.

 

[VPhantom]

Are there any backtesting software programs or online that you guys would recommend for a discretionary trader.

I don't need anything fancy just something that is literally place trade, click play and watch 3 days of price moves in one second.

It's for Windows, but I use Sierra Chart for exactly that. You can play with the replay speed at will, it's very lean on resources, and its trade simulator includes decent slippage (mostly useful for shorter timeframes). Once you're into a position, there are also handy "skip ahead" shortcuts to get to the end result faster. I haven't used it for daily bars but I assume its time controls scale up to that.

On the down side, SC has a steep initial learning curve because of the way its user interface is designed (or rather, not designed). It's what keeps the price and resources down though, so I powered through 2-3 days of just getting comfortable with it and now that it's done I'm glad I made the effort.

For me personally, I find that either small size live or simulation on past data yield about the same psychology, so I don't need to trade live when I'm developing a new idea, which is a big time saver thanks to accelerated replays. For some people, simulations "don't matter" as much and they take different decisions though, so watch out for that.

 

[dbphoenix]

I think you have to stop using the term "back test" for discretionary trading if you want to avoid confusion. In most peoples language a back test is for systems only. In a back test you take the past, and say what would system X have done in a given situation. You then see how things work out. Obviously this is repeatable over many different situations, and when you repeat the process on the same data you would get the same result.

What you're talking about is essentially "practice trading", but not in real time. You get the past and say what would I have done in that situation. You then see how things worked out.

Congrats on digging out of your drawdown BTW.

Traders were backtesting long before computers were available to them. That some reserve the term today for computerized backtesting doesn't mean that the definition has changed, anymore than the definition of "technical analysis" has changed simply because most now define it as having to do with indicators and geometric patterns.

Backtests completed via a computer program are useless to the discretionary trader as he has to see the chart in order to form even preliminary hypotheses regarding what might work and what probably won't. Discretionary trading via the results of a computer-generated backtest leads to all sorts of unrealistic expectations.

 

[VPhantom]

Discretionary trading via the results of a computer-generated backtest leads to all sorts of unrealistic expectations.

Hm, sorry I thought he meant backtest as in playback, not automation?

 

[globalarbtrader]

Traders were backtesting long before computers were available to them. That some reserve the term today for computerized backtesting doesn't mean that the definition has changed, anymore than the definition of "technical analysis" has changed simply because most now define it as having to do with indicators and geometric patterns.

That may be so, but if you google "backtest" the overwhelming majority of definitions refer to testing systematic strategies. So I stand by my statement that "In most peoples language a back test is for systems only."