Hi Everyone,
Suppose I’m examining a chart pattern. Factor A shows a 65% probability of price hitting take profit, and factor B shows an 80% probability. What we’ve got are two good probabilities working together. However, if I multiply the two, the resulting probability would be less than 65%, and that can’t be correct. What formula would I use to calculate probability in such a case?
Math brains only, please.
Thank you,
Norm
Hello, Norm
• What you’re calling “probabilities” are really just opinions.
In common parlance, we frequently hear a person assign a so-called “probability” to a particular outcome; but, that person is, in fact, merely expressing the strength of his/her opinion on a scale of 1 to 100.
[I]Such a quantified opinion has no mathematical validity,[/I] and mathematically manipulating the number assigned to that opinion produces a meaningless result.
• Second point: one outcome cannot (mathematically) have two equally valid probabilities. So, the premise of your question – that the same outcome has a 65% probability and an 80% probability – is a false premise.
• Third point, multiplying the two “probabilities” in your example implies that –
if A occurs, then there is a 65% probability that your TP will be hit
however, B says that there is only an 80% probability that A will occur.
That’s called conditional probability, but obviously it [I]does not apply[/I] to your example.
Clearly, [I]multiplying two opinions together[/I] is meaningless.
• Last, if you believe that A and B are equally sound opinions, you can simply average them and derive a sort of consensus of A and B. But, again, realize that the opinions you are averaging do not have the force of mathematical probabilities.
.
I think there is good sense in what Clint says, and the basic message is that layering factors in this way is not going to be productive. Its hard to think of where this could be practically useful in trading but maybe you have an example in mind?
The actual formula if you want to press on would be:
combined probability = 100(65/100 x 80/100) = 52%
You need to [U]average[/U] the two, Norm, rather than multiplying the two. The average (in this case 72.5%), however, assumes that equal confidence is being given to each of the two factors. If you have slightly more confidence in one factor than in the other, then the “average” should be shifted slightly in the direction of that factor.
Hmmmmmm…
I hear you. I think I understand why you’re hmmmmmming.
But Norm specified in the OP that the 65% and 80% figures provided are each an assessment of the probability of an individual trade reaching its target, according to two different methods of estimating that; so I’m sticking to my answer just above and not complicating or qualifying it. 
Though I have confessed in my previous posts to be practically an FX super-loser but have this to say:
“A” says probability of tossing a coin is 50% (if tossed enough), so does "B"
but it stays 50% naturally so why multiply. Averaging is ok here but not sure in other cases.
The probability of tossing a coin and getting heads is, yes, 50%. But the probability of getting two consecutive heads is 100(50/100 x 50/100) = 25%. It cannot be equally probable I would get two consecutive heads as just one, this is less likely.
I am not sure about sequences and will try it tonight!
possible sequences of tails heads could be more complicated.
Hi Everyone,
Let me give you an example: The Autochartist Visualizer at this address, Autochartist Visualizer | Trade Opportunities | Autochartist Visualizer API | OANDA, gives a 71% chance of an ongoing Bullish Channel Down hitting what they have identified as the target. Let’s call that factor A. Now, let’s call this factor B: Kathy Lien, in her book, The Little Book of Currency Trading, p. 114, shows that under circumstances in a particular example that she gives, 8 of 11 cases continue in a trend in her Double Bollinger Band method. That’s 73%. Of course, her example doesn’t apply specifically to the Autochartist pattern, but suppose it did.
I realize that each of these, the % and the 8 of 11 is, as applied to any particular case, perhaps a best guess or best estimate or rule of thumb, and I was looking for a formula to arrive at my own personal best guess or estimate or rule of thumb for any particular case. However, after I started the thread, I realized it was probably best to average the estimates, as some of you have suggested, and that seems to be the case now unless someone could show me otherwise.
To change the subject completely: I’ve reached a new stage in my own personal odyssey. Until now, it’s primarily been studying forex theory and strategies, and now it all seems to be coming together where my mind is automatically combining principles and techniques to form my own personal strategies, and then testing them, which I’ve only just begun to do seriously - a shift, on the whole, from study as a priority to testing as a priority. So, for those of you (two of you, in particular) who have taken an interest in my progress, thank you - and if you believe in prayer or whatever you consider to be its equivalent, please pray for me - or whatever you consider to be its equivalent.
Thank you all for your input - and happy and prosperous trading!
Norm
Its not to do with heads then heads, or heads then tails, its getting your calls right twice in a row.
Well I think we can use the formula of full probability as both factors have 1/2 chance to affect the outcome so it is
P(A|H1)+P(A|H2)=1/20.65+1/20.85=0.75 or 75%
Good try, but the figures were actually 65% and 80%, so the answer’s 72.5% as stated on the previous page (I’m not trying to be rude to you, Profitbaby, and I’m as pleased as anyone that you’re back here following your ban, but sometimes it does help to read the thread before you reply, you know? 
)
It is just the average of the factors as lexys pointed out.
(65% + 80%) / 2 = 72.5%
Although I don’t see the value in knowing that.
Hi ClarkFX,
Of course, these can’t be really accurate, but it’s like when I dodged traffic when I was a kid. If I had a pretty good idea that I could get to the other side without getting killed, I would go for it - and believe it or not, I didn’t get killed even once.
I didn’t mean to imply that one should do the actual math, but if we had a good idea of what the outcome is likely to be on a specific wisely-managed trade we’d be in a better position to decide whether to take it or not.
Norm
I, by the way, love this comment!
Well its not always the average value. One factor may appear in 1 from 100 cases and give 98% success signal while other factor may appear in 50 from 100 cases with 50% success signal. In this case full probability formula is applied instead.
[QUOTE=Jezzode;804711 “Quote Originally Posted by NormanA View Post
It’s like when I dodged traffic when I was a kid. If I had a pretty good idea that I could get to the other side without getting killed, I would go for it.”
Read more: 301 Moved Permanently
" ][B] I, by the way, love this comment![/B][/QUOTE]
Yeah, kind of like forex, isn’t it??
Best wishes,
Norm
If A and B are not dependant on the outcome of each other, then you would add the probabilities and divide by the number of probabilities, so ((65+80)/2) means you have a 72.5% chance of getting A OR B correct.
If the outcome of B is dependant on the success of A, then you have a 65% chance of getting A correct and then a 52% chance of getting B correct. This is ((65*80)/100) meaning you have a 52% chance of getting A AND B correct.
Further, lets say you have A as 65% and B as 80% and C as 20%, and they are all dependant then it’s ((((65*80)/100)*20)/100) or 10.4% chance of getting A, B, AND C correct. If they are not dependant, then it’s ((65+80+20)/3) or 55% chance of getting 1 of the 3 correct.
Probably over simplified, but hope it helps.