One of Pal’s more useful messages. I must say I agree with him. The problem with with Sharpe ratio is indeed that it penalize downside as well as upside volatility.
For example if you have two systems:
One system trade fairly infrequently, but is very accurate which results in flat periods, followed by upside jumps on the equity curve.
The second system have wild swings up and down.
Both will result in a low sharpe ratio, while the first one is a much better system, since all upside volatility is to the upside.
A system like that is very common under long term trend trading systems. These systems typically have many small losses followed by really big wins.
Regards
- Fanus
Ewart
You can already sort systems by number of trades. Do any search and just sort by number of trades.
I know you are going to get upset and tell me that you can make suggestions to Matthew all you want. See this as my request NOT to have a feauture like this.
To create a list like this, can create the perception that highly active trades are better and will result in these systems getting more exposure on the site, resulting in more views. From my viewpoint, why should a system which made more trades than mine last week get more exposure than mine?
If this is important for someone to see the number of trades per system, this is easy from them to do so on any of the search screen, by sorting according to number of trades.
Regards
- Fanus
Pal
Sometimes you tend to be too theoretical. If you have a concrete suggestion on how to better relate the profits to the risks then do it please here but concrete and practically and do not refer to expectancy articles. Subscribers need the real results and not expectancies. Please give us the calculations. Good systems have high Sharpe ratios and bad ones low. Point period. All the rest is theory. And if the C2’ subscribers were Buffett, Soros, and Robertson would not need to subscribe to C2
Regards
Sorry but you are wrong
If a system waits years in order to make an upward jump then its a bad system and it’s absolutely correct that it gets penalized by the Sharpe Ratio. But it makes sense that you get worried about the Sharpe Ratio. If you want to offer a good system then you need a high Sharpe ratio. All the rest are excuses for the ones that can’t do that
Regards
A system like that is very common under long term trend trading systems. These systems typically have many small losses followed by really big wins.
I assume that by saying “many small losses”, you mean actual “realized” losses as shown in a closed-equity curve which are not the same as “unrealized” losses which are shown in an open-equity curve as at C2 because a long-term trend following system, may have large swings in open-equity during the trade (if options are not used as an hedge), but not necessarily in closed-equity. That is why it is important to look at both the open-equity and closed-equity curve to get a complete picture of the performance of the system.
Most traders concentrate heavily on the actual entry and exit strategy, and then perform a crude position sizing technique such as trading 10 lots for every $10,000 in equity or something similar.
We all know that individual trades have different outcomes, consisting of profits that result from the entry/exit strategy and losses that result from stops (if the trade is not hedged with options.) A history of closed-out trades contains all the information needed to analyze the system’s performance and develop a position sizing algorithm. We can calculate the equity after each trade, as well drawdown, peak equity, maximum drawdown, net profit, return on equity, expectancy score, etc.
This will also answer the question, “how many shares or contracts or lots to trade?” because, even though, we know that the history of trades will not repeat, we can assume that the future will yield a series of trades with similar characteristics in terms of profit, loss, percent profitable, and so on. In other words, if we have confidence in the robustness of our strategy we can be confident that a future series of trades will be statistically similar to past history.
Out-of-sample data is essential for system validation. Robustness (a term used to describe a system or method that works under many market conditions), rather than peak performance, is the key to a useful system - Kaufman, Trading Systems & Methods"
In general I have a rule not replying to posts when someone is not willing to provide their name, but I will make an exception for you as you don’t appear to understand how Sharpe Ratio work.
Sharpe ratio basically calculate how volatile an equity curve is and was designed to work on annualized returns. It make no distinction between postive and negative moves in the equity curve.
Many systems are designed to exploit volatility. A system like that will have lots of small losses with big winners inbetween. This will result in periods of fairly flat, to slightly down slopes on the equity curve followed by big jumps to the upside. Sharpe ratio will penalized these big jumps the same as for a system which have big jumps like that up AND down.
To use your own example of a system waiting years before jumping on a trade, if the Sharpe Ratio is calculated correctly, a system like that will actually have a good Sharpe Ratio if the annualized return is higher than risk free return. Please make sure you have a thorough understanding of something before you make statements like “sorry but you are wrong”.
The biggest problem on C2 with regard to Sharpe Ratio is that Sharpe Ratio was designed to be calculated on anualized returns over many years. Calculating it over periods less than yet, will results in Sharpe Ratios jumping up and down constantly and is really not very reliable.
Regards
- Fanus
Quote:
Good systems have high Sharpe ratios and bad ones low. Point period.
This is so wrong and a statement from someone who maybe know how Sharpe Ratio gets calculated, but have no understanding what it really means. As was explained before by other people. Positive volatility is good. If a system exploit positive volatility and avoid negative volatility, this will still be penalized just the same by Sharpe Ratio.
If you have taken the time to read the article Pal posted, you would understand. You obviously didn’t even look at the article since the formula for expectancy is posted in the top of the article.
On that article is a good example:
System A returns 0.001% greater than the risk-free interest rate with zero drawdowns, and perfect consistency.
System B returns 60% per year on your account with modest 10% drawdowns.
System A will have an infinitely high Sharpe Ratio and System B will obviously be much lower.
Based on what you say is that System B is bad and System A is good. I am sorry to say, but YOU are wrong. Period.
Please open your mind a little bit and take a few moments to understand someone else’s point also. This is a clear example of you not even bothering to read about Expectancy and don’t know what it is, but just say it is wrong and Sharpe is better.
Chris
Dear Professional Traders:
My 2 cents for your review…
This is a proposal for a Score to rank all systems:
We could have something called a C2 Score which will arrive at a single metric that encompasses profitability, exposure (efficiency) and risk. C2 Score could be annualized, consequently results that span time periods of different lengths may be compared.
Exposure % is the total market exposure of the trading system: Exposure is calculated on a day by day / bar by bar basis. The Exposure metric measures the actual area of the portfolio equity that was exposed to the market.
The Annualized Risk Adjusted Return (RAR), which is simply the Annual Percentage Return, or APR, divided by percent Exposure. For profitable systems, the RAR is adjusted by multiplying the sum of 1 plus the Max Drawdown Percent (a negative number). Consequently, profitable but risky systems, which typically have periods with large drawdowns, will have a lower score than an equally profitable system that assumes less risk.
In summary, the formula for C2 Score could be:
xRAR = APR/Exposure
where xRAR is the annualized RAR.
If xRAR is greater than 0 (profitable) then
C2 Score = xRAR × ( 1 + MaxDrawDownPct )
For example, with an APR of 8.0, an Exposure of 25%, and a Max Drawdown of -15.0%, the C2 Score is:
( 8.0 / 0.25 ) ( 1 - 0.15 ) = 32 0.85 = 27.20
Non-profitable systems result in a negative C2 Score. The formula, which changes for negative RAR, more heavily penalizes non-profitable systems for their inability to overcome drawdown.
Consequently, if xRAR is less than 0 (not profitable) then
C2 Score = xRAR × ( 1 + Abs( MaxDrawDownPct ) )
Taking the same example with an APR of -8.0, we obtain a C2 Score as follows:
( -8.0 / 0.25 ) ( 1 + 0.15 ) = -32 1.15 = -36.80
This score, could be used irrespective of timeframe (hourly or annual etc), number of trades, age of the system (started 2 days ago), Dear Professional Traders:
My 2 cents for your review…
This is a proposal for a Score to rank all systems:
We could have something called a C2 Score which will arrive at a single metric that encompasses profitability, exposure (efficiency) and risk. C2 Score could be annualized, consequently results that span time periods of different lengths may be compared.
Exposure % is the total market exposure of the trading system: Exposure is calculated on a day by day basis. The Exposure metric measures the actual area of the portfolio equity that was exposed to the market.
The Annualized Risk Adjusted Return (RAR), which is simply the Annual Percentage Return, or APR, divided by percent Exposure. For profitable systems, the RAR is adjusted by multiplying the sum of 1 plus the Max Drawdown Percent (a negative number). Consequently, profitable but risky systems, which typically have periods with large drawdowns, will have a lower score than an equally profitable system that assumes less risk.
In summary, the formula for C2 Score could be:
xRAR = APR/Exposure
where xRAR is the annualized RAR.
If xRAR is greater than 0 (profitable) then
C2 Score = xRAR × ( 1 + MaxDrawDownPct )
For example, with an APR of 8.0, an Exposure of 25%, and a Max Drawdown of -15.0%, the C2 Score is:
( 8.0 / 0.25 ) ( 1 - 0.15 ) = 32 0.85 = 27.20
Non-profitable systems result in a negative C2 Score. The formula, which changes for negative RAR, more heavily penalizes non-profitable systems for their inability to overcome drawdown.
Consequently, if xRAR is less than 0 (not profitable) then
C2Score = xRAR × ( 1 + Abs( MaxDrawDownPct ) )
Taking the same example with an APR of -8.0, we obtain a C2 Score as follows:
( -8.0 / 0.25 ) ( 1 + 0.15 ) = -32 1.15 = -36.80
This score, could be used irrespective of timeframe (Dear Professional Traders:
My 2 cents for your review…
This is a proposal for a Score to rank all systems:
We could have something called a C2 Score which will arrive at a single metric that encompasses profitability, exposure (efficiency) and risk. C2 Score could be annualized, consequently results that span time periods of different lengths may be compared.
Exposure % is the total market exposure of the trading system: Exposure is calculated on a day by day basis. The Exposure metric measures the actual area of the portfolio equity that was exposed to the market.
The Annualized Risk Adjusted Return (RAR), which is simply the Annual Percentage Return, or APR, divided by percent Exposure. For profitable systems, the RAR is adjusted by multiplying the sum of 1 plus the Max Drawdown Percent (a negative number). Consequently, profitable but risky systems, which typically have periods with large drawdowns, will have a lower score than an equally profitable system that assumes less risk.
In summary, the formula for C2 Score could be:
xRAR = APR/Exposure
where xRAR is the annualized RAR.
If xRAR is greater than 0 (profitable) then
C2 Score = xRAR × ( 1 + MaxDrawDownPct )
For example, with an APR of 8.0, an Exposure of 25%, and a Max Drawdown of -15.0%, the C2 Score is:
( 8.0 / 0.25 ) ( 1 - 0.15 ) = 32 0.85 = 27.20
Non-profitable systems result in a negative C2 Score. The formula, which changes for negative RAR, more heavily penalizes non-profitable systems for their inability to overcome drawdown.
Consequently, if xRAR is less than 0 (not profitable) then
C2Score = xRAR × ( 1 + Abs( MaxDrawDownPct ) )
Taking the same example with an APR of -8.0, we obtain a C2 Score as follows:
( -8.0 / 0.25 ) ( 1 + 0.15 ) = -32 1.15 = -36.80
Systems with the best score would be the systems with the greatest return for a given drawdown and with given exposure and it is irrespective of timeframe( the length the system is held - seconds or years), number of trades, average win/loss per trade, age of system, sharp ratios or win-loss ratios.
Pardon my lengthy post.
Kind Regards and Good Luck
The keys to judging a system is its Expectancy and Profit Factor (W/L Ratio). Whether they are the ones a system should be optimized for is a moot point. As Alex Matulich says, “all the Sharpe ratio does is measure consistency. True, that’s one element of merit, but certainly not the whole picture. Using it to determine the merit of a whole trading strategy results in completely erroneous and subjective evaluations.”
>And if the C2’ subscribers were Buffett, Soros, and Robertson they would not need to subscribe to C2
The subscribers to C2 may not be those market wizards who have achieved those real results (in billions), but that does not mean they can’t use those same strategies. If we have confidence in the robustness of our strategy, then we can be confident that a future series of trades will be statistically similar to past history, (so that we can aim for the stars, and atleast we may reach the sky.) If the strategy is not robust, adapt or abandon it for a better strategy.
My Apologies. My earlier post resulted in a “double” cut and paste. Reposting again without that.
also, I would be extremely remiss if I did not say, that I am not the originator of this formula and it has been adapted for this post.
=============================================
Dear Professional Traders:
My 2 cents for your review…
This is a proposal for a Score to rank all systems:
We could have something called a C2 Score which will arrive at a single metric that encompasses profitability, exposure (efficiency) and risk. C2 Score could be annualized, consequently results that span time periods of different lengths may be compared.
Exposure % is the total market exposure of the trading system: Exposure is calculated on a day by day / bar by bar basis. The Exposure metric measures the actual area of the portfolio equity that was exposed to the market.
The Annualized Risk Adjusted Return (RAR), which is simply the Annual Percentage Return, or APR, divided by percent Exposure. For profitable systems, the RAR is adjusted by multiplying the sum of 1 plus the Max Drawdown Percent (a negative number). Consequently, profitable but risky systems, which typically have periods with large drawdowns, will have a lower score than an equally profitable system that assumes less risk.
In summary, the formula for C2 Score could be:
xRAR = APR/Exposure
where xRAR is the annualized RAR.
If xRAR is greater than 0 (profitable) then
C2 Score = xRAR × ( 1 + MaxDrawDownPct )
For example, with an APR of 8.0, an Exposure of 25%, and a Max Drawdown of -15.0%, the C2 Score is:
( 8.0 / 0.25 ) ( 1 - 0.15 ) = 32 0.85 = 27.20
Non-profitable systems result in a negative C2 Score. The formula, which changes for negative RAR, more heavily penalizes non-profitable systems for their inability to overcome drawdown.
Consequently, if xRAR is less than 0 (not profitable) then
C2 Score = xRAR × ( 1 + Abs( MaxDrawDownPct ) )
Taking the same example with an APR of -8.0, we obtain a C2 Score as follows:
( -8.0 / 0.25 ) ( 1 + 0.15 ) = -32 1.15 = -36.80
Systems with the best score would be the systems with the greatest return for a given drawdown and with given exposure and it is irrespective of timeframe( the length the system is held - seconds or years), number of trades, average win/loss per trade, age of system, sharp ratios or win-loss ratios.
Pardon my lengthy post.
Kind Regards and Good Luck
Ok. I’ll get upset and tell you go fly a kite, since you know that’s what I’ll likely do. With that aside, my suggestion for a most active category in Hot List is not simply about systems that execute the most trades. Read it again.
You said:“From my viewpoint, why should a system which made more trades than mine last week get more exposure than mine?”.
The point that I make is that systems that make the most trades and also have a comparatively high win %, demonstrate a trend for consistent and continued profitability. Secondly, the best always gets highlighted. If you have a business you always put your best product in the front display window for everyone to see.
Thanks for sharing this. It does take into account another aspect of risk, which is the market exposure, but is based on the premise that there is an inherent risk any time a position is established, ignoring the fact that options used in lieu of stops are designed precisely to minimize it.
Following this logic, the proposed C2 score would be higher (all other variables being equal) for a system that spends less time in the market, but this would severly penalize an optimal buy-and-hold approach, however, it controls drawdowns by hedging.
Just to make this discussion more interesting, here are some more measures described:
http://support.pertrac2000.com/statbody.asp
The point that I make is that systems that make the most trades and also have a comparatively high win %, demonstrate a trend for consistent and continued profitability.
Close, but not entirely accurate. I would rephrase it as follows: "systems that make the most trades (even if entries are random like in a coin toss) and also have a comparatively high expectancy, demonstrate a trend for consistent and continued profitability. " Hence the case for an Expectancy Score. %wins, whatever this number is, is irrelevant. Profit Factor (W/L Ratio) is also relevant as it is highly correlated to expectancy. Net profit or return and maximum drawdown which are determined by position sizing are equally relevant.
Also, for those who are interested, I should explain when I say that “systems that make the most trades…” means systems that takes more positions in the portfolio and trades infrequently in the context of a long-term approach, those that trades fewer positions more frequently (compared to a long-term strategy) in a med-term approach and those that trades still fewer positions, faster, in a short-term approach and those that trades very few positions, rapidly, in an ultra short-term approach.
Chris
The vendor of future pro is absolutely correct when he mentions that many of C2’s “best systems” are crappy ones. And in order to call it by name these are the ones with:
1) low Realism index
2) low P/L per unit
3) very erratic equity curve
4) low sharpe ratio
This is what helps the less experienced subscriber to be protected from them and all the rest is done by vendors offering exactly these crappy systems
Regards
Generally it would be better if you would concentrate your efforts in offering a good system instead of crappy ones. Then we could see your statements are confirmed or not.
Regards
Chris
And if you are so succesfull in correctly interpreting these ratios why don’t you offer a good system for the subscribers. You involve you in so many Forums without offering a system. If this is not the case then please correct me
Regards
Thank you for your suggestion.
This seems to be a very good proposal because it takes into account the profitability, the risk and the exposure. If it’s easy for Matthew to implement then it would be one of the most objective and fair ratios.
I thought we were discussing Sharpe Ratio? If you don’t have enough understanding about a topic in order to contribute something worthwhile and stay on the topic, this would be more beneficial for everyone if you keep quiet. Personal attacks while hiding behind anonymousity does not have any value.
The performance of my systems have nothing to do with how Sharpe Ratio gets calculated. The performance of my systems will not prove, or disprove the advantages/disadvantages of Sharpe Ratio.
A post like above does not contribute anything, other than showing your true character which explain why you post anonymously.
Regards
- Fanus
Fanus
The posts before are done for subscribers to protect them from crappy systems and not to explain them theory. If you understand the theory correctly then try to apply it to your systems and not to irritate subscribers.
