Design a BAD trading system

[meanreversion]

I think the money management thing is the best response so far. Either bet 0.01 pct of your account per trade, or 20 pct, and it's not going to work, rite?

The other good response was to trade a lot, that's probably going to hurt over time as well.

 

[arabianights]

So on the money, take one pair, enter kinda randomly based on a hunch off a low TF; it's going up, it looks like it's going down, and if the MM is sound you probably won't go bust..

If money management is perfect, you won't go bust... however your account will asymptotically approach zero

 

[MrGecko]

fade me

 

[leovirgo]

fade me

I will if you post 'real' live calls

 

[Shakone]

What if you have a 95% strike rate (winning trades as a % of total) of +10pips (+) and a stop of -20...my point is that yes a positive expectancy is always preferable but strike rate too is important in this respect.

Of course. You can make money taking profits smaller than losses. But Black Swan mentioned the entry as being random/based on a hunch. 95% win rate is going to be hard to get based on just a hunch.

 

[wackypete2]

This shouldn't be too difficult. Find your entry then martingale your way to riches, err, losses. Even if you initially risk 2% of your account and you doubled up your position after every 25 pips loss most accounts won't last a week.

Peter

 

[Hotch]

Systems don't fail because they have a negative edge, they fail because they have no edge, and then the spread kills them. Hence reversing it is going to have the exact same out come.

 

[Shanghai]

Combine a system involving nuggetology and Spanish stops. That should do the trick.

 

[meanreversion]

Systems don't fail because they have a negative edge, they fail because they have no edge, and then the spread kills them. Hence reversing it is going to have the exact same out come.

Hmmm, does this imply that "edge" is a quantifiable property, or is it simply positive expectancy = edge?

Without designing a system that dies from excessive trading, how can one possibly design something with negative expectancy, I just don't think that's possible.

Maybe it boils down to timeframe. Selling 1wk OTM options might work for a couple of years before destroying the account. Over one year, +ve expectancy, over five years, -ve expectancy.

 

[Shakone]

Without designing a system that dies from excessive trading, how can one possibly design something with negative expectancy, I just don't think that's possible.

Is this what you really mean? Surely not.

 

[Shikamaroo]

the surest way to lose money is tight stops. put in all your trades with stops 7-10 points and u're well on your way

 

[Shakone]

the surest way to lose money is tight stops. put in all your trades with stops 7-10 points and u're well on your way

Disagree. That's a sure way to lose quite often, but not necessarily a sure way to lose money overall.

 

[robster970]

Short e-mini 15 mins before RTH open on Fed POMO days.

Close position approx 90 mins later.

Repeat until US QE finishes.

 

[meanreversion]

Is this what you really mean? Surely not.

Yes, although I did go on to mention it perhaps being an issue of timeframe.

Is there a system which guarantees to lose money over time, say 5yrs?

When people backtest, they look for the system which performed the best in the past.. they then trade this and the almost inevitable result is that it performs less well in realtime than in the past.

However, let's say you find a system which has been losing money for the last 5yrs.. you could put it into production and despite it being a "bad" system, the market conditions suddenly become favourable and it starts to make money.

 

[meanreversion]

Short e-mini 15 mins before RTH open on Fed POMO days.

Close position approx 90 mins later.

Repeat until US QE finishes.

LOL good one .... go on Helicopter Ben ...

 

[Hotch]

Hmmm, does this imply that "edge" is a quantifiable property, or is it simply positive expectancy = edge?

Without designing a system that dies from excessive trading, how can one possibly design something with negative expectancy, I just don't think that's possible.

Maybe it boils down to timeframe. Selling 1wk OTM options might work for a couple of years before destroying the account. Over one year, +ve expectancy, over five years, -ve expectancy.

I'm not saying it's impossible to create a system with a negative edge, I'm saying that most of the systems you'll find (alternatively, make up random rules and backtest them, it's pretty much the same thing), don't fail because they are bad, and thus reversing them works, but fail because they don't make enough money to cover comms, so if you reverse them, you still won't make enough money to cover comms.

aka, backtest trading 5/10 EMA crossover, 10 pip target, 10 pip stop, it won't make money, then , now do the same thing, but selling instead of buying, still won't make money.