Trading probabilities

[firewalker99]

For all you trades out there, I have the following dilemma:

Assume
- You've developed a consistently profitable strategy and it works. No doubt about that. So far so good.
- You've done extensive backtesting and you've never noticed a losing streak for more than 5 successive trades (let's assume this happened 5 times). You've also noticed that your maximum string of successive profitable trades was 9 and this occurred only one time.

Suppose
- You are having a winning streak this week. So far you've had 9 successive winning trades. Would you take the next trade? The question may sound easy but depending on your viewpoint (logically, intuitively,...) there are multiple answers.

Answer
1) Of course you take the trade, if it is a valid signal according to your plan. There is no reason you shouldn't take this trade. Each trade is a new opportunity and in no way related to what happened before.

2) No, you should skip the trade. You've only had one occurence of such a winning streak through all of your backtesting and it's very improbably that you would have another winning trade.

 

[wasp]

You know you should just shut up and carry on trading the plan... the plan is the plan is the plan.... good run, bad run its all the same........ Just get on with it! Like a robot churning out products in the factory... it just goes on and on and same old....

Did you, when you decided on the plan, have a strategy when you reached X good in a row or, did you think nothing of it and keep on checking?

(sorry, that sounds a bit harsh but I hope you just see my point)...



EDIT: I know my strat has the worst probabilities on the first trade of the day. Should I lie in an extra hour each day and cut it out? No, because sometimes its the best trade of the day, and probabilities state it could be the only good trade I have some days...

 

[TWI]

I'm with Wasp on this. Question should not need to be asked. As soon as you start to bring discretion into the equation you are on a slippery slope.

I should run your trade results through a monte carlo and see what are the longest winning and losing streaks then, they will be very different at the extremes of the distribution. Don't break it by trying to make it perfect.

 

[Vorbis]

I think the correct answer is 1. The second option is based on
Gamblers Fallacy

The system itself has no memory of previous results, so the chances of success are equal in each case, all other things considered.

For example, casinos make a fortune from people who increase bet size on red on a roulette wheel after a long run of blacks. In fact the mathematical chance of a black occuring are still 50/50.

 

[firewalker99]

I think the correct answer is 1. The second option is based on
Gamblers Fallacy

The system itself has no memory of previous results, so the chances of success are equal in each case, all other things considered.

For example, casinos make a fortune from people who increase bet size on red on a roulette wheel after a long run of blacks. In fact the mathematical chance of a black occuring are still 50/50.

I agree and I think your explanation is the best. Good example, I was thinking of the coin flip analogy. If I'm correct these things are called Bernoulli experiments: each occurence is independent from what happened before and chances stay the same regardless of what happened before or after. But your reference to the Gamblers Fallacy isn't entirely correct because what you made up for a strategy is not a randomly selected number of trades but it pinpoints the logical fallacies nicely.

But let's be fair, some of us would be scared of taking the next trade after they had 5 or 6 losing trades (if their system says 5 is the maximum successive losing streak), but most of us wouldn't hesitate to take the next trade if we had 5 or 6 successive winning trades...

 

[firewalker99]

You know you should just shut up and carry on trading the plan... the plan is the plan is the plan.... good run, bad run its all the same........ Just get on with it! Like a robot churning out products in the factory... it just goes on and on and same old....

Did you, when you decided on the plan, have a strategy when you reached X good in a row or, did you think nothing of it and keep on checking?

(sorry, that sounds a bit harsh but I hope you just see my point)...

I see your point and I never said I disagreed. But you should know by now I don't "just shut up"

EDIT: I know my strat has the worst probabilities on the first trade of the day. Should I lie in an extra hour each day and cut it out? No, because sometimes its the best trade of the day, and probabilities state it could be the only good trade I have some days...

That's a tricky one but you should still take it. If you have evaluated your trades and you know for instance that certain trades have a lower probability of success you could refine your system to take only the high prob ones. But the criterium for not taking the trade should be something more scientific than "because it's the first trade of the day"... Say for example, I go over all my backtesting again and find that all trades taken around the time US markets open have only a 20% success ratio than I'd be inclined to skip those.

 

[firewalker99]

Don't want to head off-topic, but another question:

Suppose you have 3 doors and behind one of those is a treasure hidden.
You may pick any of the 3 doors (let's call them A, B and C).

Suppose you picked A.
You could be right, but we don't look just yet.
Instead, we show that there is nothing behind door C (obviously if you had chosen door C then we would show one of the other doors which have nothing behind them).

Now you're given a second chance to choose between door A and B.
Would you keep your original thought and stick with A or would you change it? And why?

(don't go surfing the web to find the answer plz we all know you can find it there

 

[frugi]

Original choice A has 1/3 chance of treasure.
While B or C taken together have 2/3 chance of treasure. (1/3 or 1/3 so we add probabilities)

You are shown C and there is no treasure. Thus B now must have 2/3 chance of treasure and A still has 1/3. So you change.

There are numerous threads elsewhere on the site about this counterintuitive problem.

 

[garethb]

Frugi beat me to it

 

[firewalker99]

Original choice A has 1/3 chance of treasure.
While B or C taken together have 2/3 chance of treasure. (1/3 or 1/3 so we add probabilities)

You are shown C and there is no treasure. Thus B now must have 2/3 chance of treasure and A still has 1/3. So you change.

There are numerous threads elsewhere on the site about this counterintuitive problem.

Yeah, it's a popular one apparently...
Guess I'd better stick to the original purpose of the thread :|
I thought most easy way to explain this one was to simply make the example with 100 or 1000 doors instead of three...

 

[rols]

Don't want to head off-topic, but another question:

Suppose you have 3 doors and behind one of those is a treasure hidden.
You may pick any of the 3 doors (let's call them A, B and C).

Suppose you picked A.
You could be right, but we don't look just yet.
Instead, we show that there is nothing behind door C (obviously if you had chosen door C then we would show one of the other doors which have nothing behind them).

Now you're given a second chance to choose between door A and B.
Would you keep your original thought and stick with A or would you change it? And why?

(don't go surfing the web to find the answer plz we all know you can find it there

Don't get it.

Why not just open all the doors at once?

What's the hidden treasure? A holiday for two in Torremolinos or a three piece suite from The World of Leather perhaps. :cheesy:

 

[temptrader]

Original choice A has 1/3 chance of treasure.
While B or C taken together have 2/3 chance of treasure. (1/3 or 1/3 so we add probabilities)

You are shown C and there is no treasure. Thus B now must have 2/3 chance of treasure and A still has 1/3. So you change.

There are numerous threads elsewhere on the site about this counterintuitive problem

Frugi,

You are talking about the Goat behind the door question. Yes, it is counter intuitive and the resolution of the question revolves around what's know as a "protocol"

 

[TWI]

What if while you are going for B the person setting the task says,
"you should not pick that door but I may be telling a lie". does this change the odds and hence your choice?

 

[firewalker99]

What if while you are going for B the person setting the task says,
"you should not pick that door but I may be telling a lie". does this change the odds and hence your choice?

Well you should never let your opinion be influenced by what others say, unless they show very solid arguments, which in this case isn't the case, so I'd wouldn't influence my choice. As for influencing the odds, obviously not.

 

[daxdaytrader]

Look at it this way: It does not matter how often you toss a coin, the next time it's still 50:50 a head or a tail. But that does assume a random outcome that does not really apply in trading where patterns really are likely. However, these patterns can also assumed to be random in terms of predictability.

So the answer is basically 1.