which one would you take and why?
(sorry that I pose questions live a professor who already knows the answer)
Why favorable R:R?
- Thread starter frostengine
- Start date
which one would you take and why?
(sorry that I pose questions live a professor who already knows the answer)
In terms of the question that was asked around the definition of expectancy...I'm sure one could potentially define expectancy many different ways, however, one that is often most helpful in terms of characterizing a trading setup/system....
-------------------------------------------------------------------------
1) WHAT IS EXPECTANCY?
Expectancy is what you can "expect" to make - on average - on a given trade and is often expressed in absolute dollar terms or as a % of $ risked. For example...
Expectancy E = (% winners * Avg win) + (% losers * Avg loss)
So consider two sets of trades (discretionary or from a system):
- Set 1
Win %: 50%
Avg Win: $100
Avg Loss: $50
Expectancy of first set = E1 = (50% * $100) + (50% * -$50) = ($50) + (-$25) = $25
- Set 2
Win %: 40%
Avg Win: $150
Avg Loss: $50
Expectancy of second set = E2 = (40% * $150) + (60% * -$50) = $30
The other way to think about expectancy is as a percentage of the $ risked:
- Set 1:
Win %: 50%
Avg Win: $100 = 2R
Avg Loss: $50 = R
Avg win in terms of avg loss = 2R
Expectancy = E1 = (50% * 2R) + (50%*-R) = .5*2R - .5*R = R - .5R = .5R (50 cents for every dollar risked)
- Set 2
Win %: 40%
Avg Win: $150 = 3R
Avg Loss: $50 = R
Expectancy = E2 = (40% * 3R) + (60% * -R) = 1.2R - .6R = .6R (.60 cents for every dollar risked)
2) HOW IS EXPECTANCY CALCULATED?
As far are I know, you have to calculate expectancy either based on back testing or based on trade statistics from actual trades. You can backtest/paper trade to calculate the above statistics. Or, lets say, you have actual trading data for 6 months - you can just look at your avg win, avg loss, % win, % loss and calculate expectancy.
Also note, lets say you trade 3 setups frequently. You can calculate expectancy for each setup independently and an overall expectancy.
3) IS HIGHER EXPECTANCY ALWAYS BETTER?
It depends. What you are looking for is a positive expectancy. A negative expectancy means that over time you will have a negative equity curve. You will lose money. You may have some positive runs, but in the long term, you will lose money with a setup/system that has negative expectancy.
So, a positive expectancy is better than a negative expectancy.
however, beyond that a higher positive expectancy may or may not be better than a lower positive expectancy. Consider the two systems above. Recall that expectancy of first setup, E1 = $25 and expectancy of second setup E2 = $30.
Lets also assume that you get five setups per day of the first type and only 2 setups per day of the second type. In this case, you will make
First setup: 5 * $25 = $125 / day (on average)
Second setup = 2 * $30 = $60 / day (on average)
So even thought the expectancy of the second setup is higher, the first setup may be better since you get more chances to trade it and therefore can net higher total $ in a given time period.
So again, you want positive expectancy; beyond that it gets a bit blurry.
4) WHAT IS THE RELATIONSHIP BETWEEN RISK TO REWARD RATIO AND EXPECTANCY?
A risk to reward ratio is an important characteristic of performance but is misses another key element: the probability of winning. As the OP noted, higher risk to reward often results in lower win ratios. It becomes increasingly difficult to maintain a high risk to reward ratio while also keeping a high win percentage (that does not mean it cannot be done).
Expectancy is a useful measure in that it combines both the risk to reward and the winning percentage.
---------------------------------------------------------------------------
Of course, knowing all this still does not help create or find higher probability setups - that comes from observing and understanding price movements. This understanding merely helps characterize different setups and helps the trader think about them in a structured and helpful way to maximize the chance of being successful.
-------------------------------------------------------------------------
1) WHAT IS EXPECTANCY?
Expectancy is what you can "expect" to make - on average - on a given trade and is often expressed in absolute dollar terms or as a % of $ risked. For example...
Expectancy E = (% winners * Avg win) + (% losers * Avg loss)
So consider two sets of trades (discretionary or from a system):
- Set 1
Win %: 50%
Avg Win: $100
Avg Loss: $50
Expectancy of first set = E1 = (50% * $100) + (50% * -$50) = ($50) + (-$25) = $25
- Set 2
Win %: 40%
Avg Win: $150
Avg Loss: $50
Expectancy of second set = E2 = (40% * $150) + (60% * -$50) = $30
The other way to think about expectancy is as a percentage of the $ risked:
- Set 1:
Win %: 50%
Avg Win: $100 = 2R
Avg Loss: $50 = R
Avg win in terms of avg loss = 2R
Expectancy = E1 = (50% * 2R) + (50%*-R) = .5*2R - .5*R = R - .5R = .5R (50 cents for every dollar risked)
- Set 2
Win %: 40%
Avg Win: $150 = 3R
Avg Loss: $50 = R
Expectancy = E2 = (40% * 3R) + (60% * -R) = 1.2R - .6R = .6R (.60 cents for every dollar risked)
2) HOW IS EXPECTANCY CALCULATED?
As far are I know, you have to calculate expectancy either based on back testing or based on trade statistics from actual trades. You can backtest/paper trade to calculate the above statistics. Or, lets say, you have actual trading data for 6 months - you can just look at your avg win, avg loss, % win, % loss and calculate expectancy.
Also note, lets say you trade 3 setups frequently. You can calculate expectancy for each setup independently and an overall expectancy.
3) IS HIGHER EXPECTANCY ALWAYS BETTER?
It depends. What you are looking for is a positive expectancy. A negative expectancy means that over time you will have a negative equity curve. You will lose money. You may have some positive runs, but in the long term, you will lose money with a setup/system that has negative expectancy.
So, a positive expectancy is better than a negative expectancy.
however, beyond that a higher positive expectancy may or may not be better than a lower positive expectancy. Consider the two systems above. Recall that expectancy of first setup, E1 = $25 and expectancy of second setup E2 = $30.
Lets also assume that you get five setups per day of the first type and only 2 setups per day of the second type. In this case, you will make
First setup: 5 * $25 = $125 / day (on average)
Second setup = 2 * $30 = $60 / day (on average)
So even thought the expectancy of the second setup is higher, the first setup may be better since you get more chances to trade it and therefore can net higher total $ in a given time period.
So again, you want positive expectancy; beyond that it gets a bit blurry.
4) WHAT IS THE RELATIONSHIP BETWEEN RISK TO REWARD RATIO AND EXPECTANCY?
A risk to reward ratio is an important characteristic of performance but is misses another key element: the probability of winning. As the OP noted, higher risk to reward often results in lower win ratios. It becomes increasingly difficult to maintain a high risk to reward ratio while also keeping a high win percentage (that does not mean it cannot be done).
Expectancy is a useful measure in that it combines both the risk to reward and the winning percentage.
---------------------------------------------------------------------------
Of course, knowing all this still does not help create or find higher probability setups - that comes from observing and understanding price movements. This understanding merely helps characterize different setups and helps the trader think about them in a structured and helpful way to maximize the chance of being successful.
Well, I wouldn't say I "know" the answer, so if you do know and could explain your reasoning then I'd certainly be interested
In a perfect world I would prefer a system (or combination of different systems) that takes a large number of trades, with R:R of at least 1:1, with at least 60% accuracy. Small losses, with more frequent and slightly larger gains. This would produce a smooth, rising equity curve. Unfortunately the world isn't perfect, and I haven't yet identified such a system. Any system that relies on extreme values (like 1:10 risk:reward or 90% accuracy) seems to me inherently unstable.
What I can say is that I personally have found it impossible to achieve high accuracy with my entries. I would say 50% of my trades move less than 2 ES points in my favor and 25% (half the above) move less than one point. If consistent, long term profitability is in the cards for me it's going to be from the outlier trades and monster moves where I can reasonably make 10x or 15x risk. These do appear regularly in the markets, the problem is getting in and staying in while keeping the frequency and size of your losses manageable.
Interesting question. Here's my thinking....from a purely conceptual perspective, it should not matter. However, from a practical standpoint, there are a number of other, often inter-related, considerations:
a) # of times you get to trade each setup. If two setups/systems have equal expectancy, than the one that you can trade more often in a given period will yield higher $ / period
b) Psychology (I): It is very tough for most people to trade a system with a 10% accuracy - you can have an unbearably large numebr of consecutive losers. Is your system not working or just going through an expected losing streak.
c) Psychology (II): The equity curve on a system with 10% win ratio will tends to be very erratic.This is similar to (b) but I feel a different aspect that makes it tough to trade such as system vs. a system witha higher win raito but same expectancy
d) Friction: Often execution (order fills, slippage) eat into avg win numbers and exagerate avg losses - the higher the avg. win, the easier it is to translate the setup/system from concept to reality (gross generality. I know).
e) Scaling: Systems with known large winners and low winning % allow you to scale when you are right and can often lead to overall greater wealth creation. However, again, points (b) and (c) make these systems these systems almost impossible for most (retail) traders. )Although, in the strictest sense, a system that scales is now not really the original system anyway.)
So my answer is that it depends. You have to find what fits your trading style- knowing that in some cases you may be trading off $ for comfort.
You know, your reponse was kind of rude.
1) Actually, if you look back at his response to the above quote, he agreed that the way I had first said it was not quantifiable was correct. He then went on to say "probability of the expectancy" which I still feel is pretty darn vague and subjective and kind of cracked me up - this doesn't mean I am not trying to see what he means by it - have you tried to see what I meant?
2) Look at the first post of the thread - the OP specifically said that his topic was NOT about expectancy - I simply challenged why Osorico and apparently you are trying to substitute expectancy, which has a high degree of subjectivity - like you said, "judging the profitablility" and "felt this, felt that", when that was not the aspect of trading that OP was trying to explore
3) You are making some pretty big negative assumptions about what I believe and how I trade based on no knowledge of that at all, only your own biases. This is the sign of a small, angry mind. It is not me that is pissed off, it is apparently you.
By the way, you couldn't be more wrong. I am a pure price action momentum daytrader. I use no indicators or fundamentals or system. Yes, real money. Yes, profitable.
4) I thoroughly understand expectancy, apparently more than you who can't comprehend its limitations as a stand-alone metric for trading.
In your example above, even if the trader does win 75% of the time, he will lose money or at best have lackluster profits due to a too big stop loss size, slippage, commish, etc. if those figures are not controlled to maintain the edge. In that case, a trader would have a false expectancy, woudn't they?
Or said trader will delude themselves by looking at yesterday's charts and thinking they have a robust methodology expectation when in forward realtime, they really don't, because they can't recognize said setup as it is happening, only later when the market is closed. Or they falsely pull the trigger when the setup is not there. Or don't wait for confirmation. Or any one of many ways they wreck their own plan.
Peace now, let's have a polite educational debate and not a personality clash.
Quote from Specterx:
Well, I wouldn't say I "know" the answer, so if you do know and could explain your reasoning then I'd certainly be interested
In a perfect world I would prefer a system (or combination of different systems) that takes a large number of trades, with R:R of at least 1:1, with at least 60% accuracy. Small losses, with more frequent and slightly larger gains. This would produce a smooth, rising equity curve. Unfortunately the world isn't perfect, and I haven't yet identified such a system. Any system that relies on extreme values (like 1:10 risk:reward or 90% accuracy) seems to me inherently unstable.
1. I assume that one knows how to calculate the variance.
2. The average performance of a system is LESS than the arithmetic average (the one referred to in this thread as expectancy). In fact the true return of a system is its GEORMETIC average.
3. To understand geometic average's implications, you should know that it can its logarithm can be approximated by a formal like this:
Arithmetic Average - Variance*(positive term).
What this means is that if you increase Variance, your geometric average decreases.
When the above term is zero, it means that your geometic average is 1 (meaning you do not make money neither lose even if your arithmetic average is positive). If variance is high you can even lose money. That is why traders use only a fraction of their capital to make sure they have a positive geometric average.
4. Now let us address the main question. If you have two systems, the one with higher variance will lead to a lower goemetric return.
5. Note that if I have a system with lower variance, I can always increase its variance by increasing leverage. The leverage can be increased so that the variance of system 1 have the same var as the system2 that had a high variance, but with the benefit of a high expected return component, and therefore a higher geometric return.
The above demonstrates that for same expected average return, pick the one with lower variance.
Lower variance also has the added psychological part as the equity curve looks smoother ( assuming the same leverage for both system).
You have put your finger on a key question in trading. I have developed a problem which I call "The Monkey Trading Problem" to prove the exact nature of what you have encountered, and how to deal with it.
Please search for it, and post the link here. Thank you (It is the education or professional forum). Posted on December 11 (early hours of the morning).
Continue your good thinking. You are on the right path, but not quite there yet from what I understand from your post.
This is incorrect, and other have made the same mistakes in other posts. There is another post that tried to correct this, but somehow people seem not to have noticed it.
Correct: the risk part (not the R/R ratio) that affects the probs.