High hit rate vs big R:R - Which one do you prefer and why? | Page 6 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Which one do you prefer: Metatrader (MQL) or cTrader (cAlgo)? 17 replies

Bar chart vs Candle stick which one do you prefer? 6 replies

Risk/Reward and % win... Which do you prefer? 8 replies

Long term or short term which one u prefer? 5 replies

which one you prefer? 6 replies

Attachments: High hit rate vs big R:R - Which one do you prefer and why?
Exit Attachments

High hit rate vs big R:R - Which one do you prefer and why?

  • #101
  • Oct 11, 2015 2:05am Oct 11, 2015 2:05am
Quoting Copernicus
{quote} No probs. If you want to determine what Van Tharpe uses as a basis to assess system performance (System Quality Number SQN) then use the following formula over consecutive n=100 sample sizes for as long as you can go and then average the total result to get a total result. The SQN is a very effective relative measure to assess system robusteness in tandem with performance and then utilise this to optimise position sizing to really ram results home :-) Embedded in the formula is expectancy plus sufficient sample size plus volatility of returns....
Thx I will look in to it..... But i think the best formula to determine you system is ...you know what i am going to say.... Green pips Copernicus
  • #102
  • Edited 2:57am Oct 11, 2015 2:38am | Edited 2:57am
Quoting Copernicus
{quote} SQN = Squareroot (#Trades) * Avg R/Std Dev R {image}
I was playing around with SQN recently and after a while realized that if you assume your risk risk per trade is simply your average loss (and I don't see why you wouldn't do this in spite of what Tharp says) then the equation comes down to Sharpe ratio * sqrt(trades) or Sharpe * 10 when trades > 100. I don't want to criticize SQN because it is based on a valid mathematical model (Sharpe ratio), but simply point out that SQN is a scaled Sharpe ratio. In this case you would assume a risk-free rate of 0 for the Sharpe ratio calculation: Avg return / Std Dev of returns.

Just another (of many) examples of a trading book writer cashing in on an existing mathematical formula by adding a simple tweak, repackaging, or obfuscating the formula. It's funny because in the book "Definitive Guide to Position Sizing" Tharp criticizes Ralph Vince for doing essentially the same thing with regards to cribbing Kelly's ratio and creating Vince's optimal f. Birds of a feather, I suppose.
  • #103
  • Oct 11, 2015 2:43am Oct 11, 2015 2:43am
Quoting FXEZ
{quote} I was playing around with SQN recently and after a while realized that if you assume your risk risk per trade is simply your average loss (and I don't see why you wouldn't do this in spite of what Tharp says) then the equation comes down to Sharpe ratio * sqrt(trades) or Sharpe * 10 when trades > 100. I don't want to criticize SQN because it is based on a valid mathematical model (Sharpe ratio), but simply point out that SQN is a scaled Sharpe ratio. In this case you would assume a risk-free rate of 0 for Sharpe. Just another (of many) examples...
Nice rationale. They have to make money some how however I s'pose. I must admit that I have a fond affection for Van Tharpe though despite your conclusion which is probably exactly how he derived SQN as he at least places a traders foot on solid statistical ground. :-)
  • #104
  • Oct 11, 2015 3:12am Oct 11, 2015 3:12am
Quoting Copernicus
{quote} Nice rationale. They have to make money some how however I s'pose. I must admit that I have a fond affection for Van Tharpe though despite your conclusion which is probably exactly how he derived SQN as he at least places a traders foot on solid statistical ground. :-)
True, the ratio is sound and the idea of scaling Sharpe's ratio by trades is in interesting innovation. Perhaps the scaling was inspired by Kelly's expected long term growth rato, which plots a single point on Markowitz's efficient frontier and is another of my favorite ratios.
Inserted Code
(Avg return ^2 * trades) / (2* variance of returns)
(citation Paul Wilmott on Quantitative Finance, 2nd Edition, Chapter 17, formula 17.2).
  • #105
  • Oct 11, 2015 4:46am Oct 11, 2015 4:46am
A couple of people have already said it.

Try trading high win rate systems, after a while you realise that they don't work.

High reward systems, or following the trend does work.

Like others have also said, the reward is not knowable before you enter the trade, tharpe's expectancy looks great in hindsight but not knowable in advance.

So the motto of cut winners and run losers is the only thing that actually works.
  • #106
  • Oct 11, 2015 5:56am Oct 11, 2015 5:56am
Quoting alphaomega
I have a choice I always prefer the high win rate. And by the way 65% with 1:1 definitely is possible and achievable.. Many people have done it (even better than that). Just one famous example - Martin S. Schwartz. This guy had around 70% win rate, RR 1:1 or higher and never had more than 3-4% drawdown. He turned 50K into 20 mil. in a few years back in the 80's, won several trading championships. Absolutely amazing. There are many other, less famous examples. So it is possible, but not probable for 99% of the people. You really,...
I do agree with this. But let's pretend I'm an aspiring trader who has the goal of being profitable instead of being famous. I have two choices:

- Either I try to be the new Martin S. Schwartz. Unfortnately we are no longer in the 80's. I know it will be even more diffucult for me to find such an edge.

- Or I humble down and I accept a low WR with a good RR by simply being as stupid as a seagull following a fishing boat to grab his food. Note how much the bird don't care where the boat will be going.

In the first case I can blow accounts after accounts with 1% chance or less of being successful some day. Out of 8 billions human beings twelve people have walked on the Moon. In the second non-glamourous scenario at least it works. Path of weak resistance to quote another well known trader.
No greed. No fear. Just maths.
  • #107
  • Oct 11, 2015 6:08am Oct 11, 2015 6:08am
Quoting PipMeUp
{quote} I do agree with this. But let's pretend I'm an aspiring trader who has the goal of being profitable instead of being famous. I have two choices: - Either I try to be the new Martin S. Schwartz. Unfortnately we are no longer in the 80's. I know it will be even more diffucult for me to find such an edge. - Or I humble down and I accept a low WR with a good RR by simply being as stupid as a seagull following a fishing boat to grab his food. Note how much the bird don't care where the boat will be going. In the first case I can blow accounts...
That has always been my goal in trading. To think a little as possible during the trading session. I save all my thinking for the weekend. For me, I don't care at all about the win rate. Manage your powder responsibly and the trend will come.
If you are good at something, never do it for free--Joker
  • #108
  • Oct 11, 2015 7:57am Oct 11, 2015 7:57am
Quoting hanover
My experiences have led me to much the same conclusion. Genuine edges come from somehow finding and exploiting genuine inefficiencies in markets that are generally very efficient.
I think that exploiting inefficiencies occupies the realm of arbitrage. Speculative traders tend to select markets that are efficient. The reason for the profitability (or edge) of technical trading may never be fully understood. To the outsider it just looks like randomly opening 10 trades, 9 make a loss and 1 becomes a large winner driven by a sequence of events that could never have been predicted at trade inception. I personally think that technical analysis allows one to 'read' market sentiment and identify possible occasions when a market may be primed for a move but the triggering 'X' event necessary to ignite the move is something that literally nobody can predict, even with perfect fundamental information.
Quoting Copernicus
Be aware of Atokys as well as I suspect he is another Denizen of the deep.
  • #109
  • Edited 12:55pm Oct 11, 2015 12:28pm | Edited 12:55pm
Quoting Atokys
I think that exploiting inefficiencies occupies the realm of arbitrage..........I personally think that technical analysis allows one to 'read' market sentiment and identify possible occasions when a market may be primed for a move but the triggering 'X' event necessary to ignite the move is something that literally nobody can predict, even with perfect fundamental information.
Agree about the limitations of TA. I also believe that opportunities for arbitrage generally lie outside of the reach of the retail trader. That doesn't leave very much.

I recently watched a video of an interview with a successful fund manager, and he attributed his firm's superior performance record to their ability to anticipate the trend potential of different (both forex and other) markets better than their competitors.
  • #110
  • Edited 7:03am Oct 12, 2015 6:50am | Edited 7:03am
Quoting jmn5611
{quote} That has always been my goal in trading. To think a little as possible during the trading session. I save all my thinking for the weekend. For me, I don't care at all about the win rate. Manage your powder responsibly and the trend will come.
That is spot on. You simply cannot predict the market. It is adaptive, globally connected, arbitrage opportunities continuously shift within and across markets and dissipate in a blink of an eye. The market is hopelessly complex. The traditional simplifed viewpoints that surmise by theoretical assumptions or use simplified reduction are just plain wrong or comedically simplified. The 'state' of a complex system continuosly evolves and adapts just as a biological organism does by virtue of it's contents and co-dependencies. Equlibrium between nested systems and subsystems continuously shifts as arbitrage opportunities get explioted across and within markets. They key to survival in this market is risk management which leads to survival of the fittest simply by virtue that the survivors stay around long enough to exploit any arbitrage opportunity that may unpredictably be thrown their way.

But this is where 'a robust system' comes in. A robust system is not one designed for pure performance but most importantly for survival, and you actually need to delve a bit further than win rate, R:R and expectancy alone to justify a robust system. Risk is not fully featured in expectancy and you need to closely look at variability of returns in association with differring market conditions to ensure it is sufficiently robust enough to survive a storm. Unfortunately even a robust system cannot fully deal with market uncertainty but probabilities for a robust system conclude that the odds are in your favour that you are more likely to be the survivor in the system and participate in any spoils if available.
  • #111
  • Oct 12, 2015 8:14am Oct 12, 2015 8:14am
Quoting hanover
{quote} Agree about the limitations of TA. I also believe that opportunities for arbitrage generally lie outside of the reach of the retail trader. That doesn't leave very much.
I don't view the points made as limitations to TA. Trading is fundamentally about dealing with uncertainty and TA offers the best chance at success if used to quantify and manage risk.
Quote
I recently watched a video of an interview with a successful fund manager, and he attributed his firm's superior performance record to their ability to anticipate the trend potential of different (both forex and other) markets better than their competitors.
Thats probably just self promotion. There are no secrets when it comes to following trends.
  • #112
  • Oct 12, 2015 10:37pm Oct 12, 2015 10:37pm
Quoting Kong
I really don't know how to explain it in a sentence but imo to have a preference for either or is missing the point. Unless of course you are too busy or unpassionate about learning to be an actual professional in this business, then sure, pick one or the other but there's no right or wrong answer, it's just a metric applied to some end objective.
The goal of this thread is to provide information about the extremes.

As I mentioned before, there are champion traders in both camps and they are regarded as super traders. Of course, there are always those somewhere in the middle of the extremes and there exists champion traders there too but things from the extreme sides are more interesting because by default they are already extraordinary.

However, this thread is filled with opinions from the extreme sides and also those in the middle so hopefully those looking for information will be supplied by the generous knowledge shared here.


Quoting antonsatu
IT IS IRRELEVANT to compare RR and win rate.

We need both. And we can have both.
Any kind of opinion is respected and yes, you can have one of them or both according to your preference.


Quoting hanover
When looking at trader performance on forums, very high win rates are frequently the result of unsustainable MM (e.g. excessively wide or no SL; adding to open losses; martingale variants). Obviously you could potentially fix your system RR by employing a rigid SL/TP across every position, but there is no guarantee that this in itself will provide an edge.
Wisdom from the old timer. One of the rare gems


Quoting alphaomega
What I wanted to show is that between strategies with similar expectation the one with higher win rate is always superior. Also, lower expected drawdown and smoother equity curve allows the trader to take higher risk per trade relatively safely. With higher win rate, the "gambler's ruin problem" doesn't affect you too much. With low win rate not only you have to trade smaller by default, but you constantly have to reduce(progressively) the trade size, because if you don't - you will lose everything in the long run. Because these strategies spend...
I can't afford not to quote this one. Very good view from the high WR side. Fits perfectly to the description of the thread.


Quoting PipMeUp
I do agree with this. But let's pretend I'm an aspiring trader who has the goal of being profitable instead of being famous. I have two choices:

- Either I try to be the new Martin S. Schwartz. Unfortnately we are no longer in the 80's. I know it will be even more diffucult for me to find such an edge.

- Or I humble down and I accept a low WR with a good RR by simply being as stupid as a seagull following a fishing boat to grab his food. Note how much the bird don't care where the boat will be going.

In the first case I can blow accounts after...
This is exactly what this thread is all about.

People have different opinion and they also have a choice. By your guys sharing different angles to the same problem the new starters or those struggling can decide which one is good for them. They can try to become one of those rare traders on either side or settle somewhere in the middle. When they fail in their attempt they can always change side to see which one really work for them
Different/opposing ideas will make us learn and understand many things.
  • #113
  • Oct 12, 2015 11:35pm Oct 12, 2015 11:35pm
Quoting antonsatu
IT IS IRRELEVANT to compare RR and win rate. We need both. And we can have both.
RR and win rate are inversely proportional, one goes up another goes down. u cant have the best of both world.
  • #114
  • Oct 13, 2015 1:45am Oct 13, 2015 1:45am
IMO,
If an account is in profit then:
High hit rate (more than 80 % wins) = skill ; Big RR ( less than 30 % wins) = luck
  • #115
  • Oct 13, 2015 4:45am Oct 13, 2015 4:45am
Quoting Mingary
IMO, If an account is in profit then: High hit rate (more than 80 % wins) = skill ; Big RR ( less than 30 % wins) = luck
Hmm... I think this is true but only to a certain degree.

Your definition of luck apply only to new traders or those with limited library of sample. Those who are in profit using big RR technique with large quantity of data (a thousand or more trades) can't simply rely on luck to survive the losing streak periods.

What do you think?
Different/opposing ideas will make us learn and understand many things.
  • #116
  • Oct 13, 2015 6:35am Oct 13, 2015 6:35am
Quote
Big RR ( less than 30 % wins) = luck

Not necessary IMO... A characteristic of long-term trend following systems seems to frequently be a low win rate but a high RR....(this was my experience when trading equities anyway)... if the strategy has been designed to capture such long term trades and the trader is following the system correctly, then why should this only be associated with luck.
  • #117
  • Oct 13, 2015 8:12am Oct 13, 2015 8:12am
Hit rate far more important as it tells you far more in terms of money management. That is, hit rate, p, or expected hit rate in the future, p', can be used for a "best guess" estimate of the asymptotic peak of the expected growth curve as p'/2.

Other geometrically important points can thus be estimated from this, the inflection point, the reward/risk maximizing point, etc., in similar fashion with a divisor that also incorporated number ot trials (as these are germane to any points on the curve -- including the asymptotic peak, which (the divisor for best guess estimate) moves from p' up to 2 as the number of trials -> infinity . -Ralph Vince
  • #118
  • Oct 13, 2015 5:31pm Oct 13, 2015 5:31pm
Quoting rvince
Hit rate far more important as it tells you far more in terms of money management. That is, hit rate, p, or expected hit rate in the future, p', can be used for a "best guess" estimate of the asymptotic peak of the expected growth curve as p'/2. Other geometrically important points can thus be estimated from this, the inflection point, the reward/risk maximizing point, etc., in similar fashion with a divisor that also incorporated number ot trials (as these are germane to any points on the curve -- including the asymptotic peak, which (the divisor...
Hi rvince (sctarches head in sceptic incredulity).

It would be grealy appreciated if you could please delve into this a bit more in more simplistic terms so I can get on the same page as you as I would like to follow you deeper into this viewpoint as it challenges some of the opinions that guide my trading style.
  • #119
  • Oct 13, 2015 6:31pm Oct 13, 2015 6:31pm
Quoting Mingary
IMO, If an account is in profit then: High hit rate (more than 80 % wins) = skill ; Big RR ( less than 30 % wins) = luck
Regardless of the output results in terms of win rate and RR, ANY strategy that is profitable after at least 1000 trades with net profit at least 2 times bigger than max drawdown - this is serious evidence for skill. You just cannot achieve such result by pure luck. Not in a million years.
Also if someone here have similar result - don't forget the ultimate truth about trading:
PAST PERFORMANCE IS NOT INDICATIVE OF FUTURE RESULTS!
95% of traders read this warning almost on daily basis, and still don't understand what it actually means.
Markets are just another form of darwinian evolution. "Survival of the fittest" Only the top 4-5 %, most skilled and the most adaptable traders will survive in the long run. The others will be eaten by the "monster" !
Or to quote one BIG floor trader from Chicago - "The market is a whore! It's out there to fuck, as many people as it can."
  • #120
  • Oct 13, 2015 6:59pm Oct 13, 2015 6:59pm
Copernicus,

You want robustness in your position sizing with respect to future market conditions. can we stipulate that this is paramount?

That being the case (and we are only discussing the simple case of trading one component at a time here), our criterion (in the sizes we are trading in) may not be to maximize (asymptotic) long-term expected growth, but perhaps something else (e.g. maximize MAR ratio, or something like that).

We can estimate the peak of the curve as p'/2 because (if you are using the optimal f formula, which is "scaled,"; if you use the formula that satisfies the Kelly Criterion, without scaling it, you're going to be lost) the peak, in the future Q periods or trades, is going to be bound between 0 and p (or, p' which is p over the Q future periods).

Since the price one pays for missing the peak grows with Q (because where you are on the asymptotic expected growth function, which is optimal f as Q->infinity, or simply The Kelly Criterion solution, scaled properly , the the Kelly Criterion Solution is an often-improperly scaled, always asymptotic [i.e. as Q->infinity] solution, which further assumes that each Q occurs instantaneously) such that whatever point you are on the curve, G', the expected growth curve in the future, is taken to the power of Q, then just like with the calculation of Variance, where you pay a quadratic price for the farther an outlier is from the mean, similarly, you pay a price of the power of Q for however far you are from where the peak of the curve, f'{0,p'}. Thus, you want to minimize the worst-case outlier, and this is simply a matter of dividing the range that f', the future value for the asymptotic growth-optimal fraction to risk will be, which is {0,p'} so as to be in the center of that range, p'/2, which is therefore the most that an outlier -- i.e. the most you can miss the peak by.

Back to why you want to know the peak regardless of your criterion, this is because other points of geometrical significance can be determined robustly in similar manner.

I hope this explains it? I hope it answers what you are asking clarification on here. If you want to learn more about these other points of geometrical significance (not the robust means of calculating them, but the actual, mathematical means) there are two papers that you might wish to look at:

Lopez de Prado, Marcos and Vince, Ralph and Zhu, Qiji Jim, Optimal Risk Budgeting Under a Finite Investment Horizon (October 5, 2013). Available at SSRN: http://ssrn.com/abstract=2364092

and

Vince, Ralph and Zhu, Qiji Jim. Optimal Betting Sizes for the Game of Blackjack. Journal of Investment Strategies 4(4), 53-75.

-Ralph Vince
www.ralphvince.com
0 traders viewing now