How much would you risk on a GOOD trade?

[macintash]

Quote from k p:

Here is my stab at it. If its a coin toss, then you have a 50/50 chance. So if you were to ask how much do you bet on this, I think as others are saying 2% sounds about right. I'm not well trained in statistics, but I think the idea that coming up heads in a coin toss 10 or 15 times in a row is quite remote... 20 times in a row would be incredibly unlikely. So if you bet 2% of your account on a 50/50, even if wrong 20 times in a row, you have only lost 40% of your account, and who the heck thinks a coin will come up heads 20 times in a row??

Now if you are saying you have 95% probability, then lets call this a 20 sided coin, and hence only one side of the coin, or rather 5% would be the bad side, and the other 19 sides would all prove to be winners if it landed on that side. So what are the chances that it even lands 2 times in a row on that bad side? Like I said, I'm not good at the math, but if I have this right, the chance of landing on the bad side 4 times in a row which has only a 5% chance would be 0.05 raised to the power of 4 or 0.000625%. Am I doing this right???

So I would call that extremely small, and I think that to lose 4 times in a row on that bet would be extremely unlikely, and hence I would bet 25% of my capital if I had a 95% chance of for sure being a winner given a coin toss that had 20 sides as in my example. I'm not sure how you arrive at a 95% chance of being right, but if this in in fact true, then being wrong 4 times in a row would be quite unlikely and risking 25% of you capital for each bet shouldn't wipe out your account. Hmmm.. then again, your chances of winning the lotto are even more rare than that and people do win every day, so maybe forget what I said! LOL

The problem with the comparison to a 20 sided coin is the following.
1) While based on analysis/research/history it seems that the chance for profit is 95%, you might have made a mistake with the analysis (mistakes happen by everyone)
2) Markets change. Even if something is 95% today, it can be different tomorrow. A 20 sided coin is a 20 sided coin, and always stays that way.
That’s also the reason why Kelly is hard to justify. The question is where the balance is.

 

[xelite777]

Markets change.

For the last time, markets do NOT change, they go up, they go down, they go sideways, THAT'S ALL THEY HAVE BEEN DOING SINCE DAY ONE!!

Buy when they go up, short when they go down, and do nothing (or sell options) when they are going nowhere, that's all there is to it.

 

[k p]

Quote from macintash:

The problem with the comparison to a 20 sided coin is the following.
1) While based on analysis/research/history it seems that the chance for profit is 95%, you might have made a mistake with the analysis (mistakes happen by everyone)

Oh I completely agree, and that is why I used my coin idea because it is something concrete. But I hardly doubt that in trading there can be a 95% chance of something happening the way you thought it would.

 

[mercadomagico]

i make my trades riskless as possible.. low volume, tight spread, 1.00 intraday range. After a while you just know what to risk

 

[xelite777]

Quote from k p:

But I hardly doubt that in trading there can be a 95% chance of something happening the way you thought it would.

Sure you can.

Buy a $100 stock. Put your take profit at $105, no stop loss.

You will win that trade 95% of the time.

Most traders do not realize that profitability in trading has nothing to do with percentage of winning trades.

 

[k p]

Quote from xelite777:

Sure you can.

Buy a $100 stock. Put your take profit at $105, no stop loss.

You will win that trade 95% of the time.

Most traders do not realize that profitability in trading has nothing to do with percentage of winning trades.

So what you're saying is that 95% of the stocks out there will at some point be worth 5% more than they are at the present day? I don't think I can believe this. I have no idea what the actual number is, but even without a time frame, I doubt that 95% of all stocks will one day be worth 5% more than they are today. Can you back this up in any way?

 

[xelite777]

Quote from k p:

So what you're saying is that 95% of the stocks out there will at some point be worth 5% more than they are at the present day? I don't think I can believe this. I have no idea what the actual number is, but even without a time frame, I doubt that 95% of all stocks will one day be worth 5% more than they are today. Can you back this up in any way?

A simple example to illustrate:

We are playing Head or Tail for a dollar per game. We each have a $100 bankroll and we will keep playing until one of us is broke.

99% of the time, I will win at least $1 before you take my entire bankroll.

Same thing here with stocks, it's just simple mathematics.

 

[ras72]

Quote from macintash:

If you find a trade with a high probability (95% confidence rate) of making a nice profit, how much % of account value would you risk? 2%? 5%? 15%? 35%?

The question is posed in improper terms and cannot be answered without considering win and loss.

If for example the potential loss were to be 20% of the wager and potential win 40%, OBVIOUSLY, the sensible thing would be to wage the entire capital, thus risk 20% per event, in your terms. This except if there's the risk of capital dropping below a participation treshold in which case one accomodates for, say, 4 losses in a row (p=0.0005). That's an arbitrary parameter.
(Assuming no impact in relation to some utility function of time or alternative uses for the capital)

Of course the choice becomes more difficult as the profitability of your hypothetical, p=0.95, game decreases, as the potential win becomes significantly smaller than the potential gain (absolute terms R<<1). Also at some point as R << 1 the game's capital growth curves could show such high variance as to bring into consideration the risk of a lethal adverse excursion (ie. in case of a participation treshold). The profit maximizing fraction of capital, 'f', will however be less than 1 and greater than 0 only for a very limited range of 'p' and 'R' combinations resulting in a 'g' (mean geometric growth rate of capital) ~1.005; so for marginally profitable games.
More typically 'f' will be either 1 (bet everything) or 0 (don't bet).

- ras72

 

[ras72]

Errata corrige:
"Of course the choice becomes more difficult as the profitability of your hypothetical, p=0.95, game decreases, as the potential win becomes significantly smaller than the potential loss (absolute terms R<<1)..."

The bottom line is that if the game is in the small range in which the profit maximizing 'f' is between zero and one get a software to calculate it.

- ras72

 

[Swan Noir]

Please indulge my simplistic nature and take me through how you calculate that there is a 99% chance it will turn out as you say. I'm not questioning that you are correct ... the result makes sense to me in a common sense way but frankly I do not know how to set the problem up to solve the problem or others that are similar and I would like to be able to do that.

Quote from xelite777:

A simple example to illustrate:

We are playing Head or Tail for a dollar per game. We each have a $100 bankroll and we will keep playing until one of us is broke.

99% of the time, I will win at least $1 before you take my entire bankroll.

Same thing here with stocks, it's just simple mathematics.