- Joined
- 28 December 2013
- Posts
- 6,392
- Reactions
- 24,319
New traders attracted to the obvious break-out trading style become quickly disillusioned when their W% falls below 50%. They think they're doing something wrong. I've had people wonder why my long term W% isn't >50% if I'm a good trader. I was going to comment that one can increase the W% of a trend following system by taking smaller profits. Pleased to see that you realise this and you also know the cost of doing this (less profit).
so, the obvious question:
why would anyone prefer win rate over profit?
Personally, I don’t care about win rate. It doesn’t mess with my head on a day to day basis when live trading and it is certainly not high on my list when accessing systems.
I want to explore why "trend trading systems" have a low win rate.
I think your systems have done a stellar job.
I just got the email too. It’s very interesting and I’m curious to explore the outlined methodology ?Received a befitting email with regards to position sizing from nick radge (if you're subscribed to his mailing list you would of too). if you haven't here it is. A very interesting concept about distribution of risk. Anyone ever tried it? This is a straight cut and paste
There are many different ways to grow an account. Using a linear method such as Fixed Fractional position sizing will do it, albeit slowly and surely. The key to growing the account is that you need to take on more risk, and taking on more risk means exposing oneself to a higher drawdown.
Drawdowns on paper are easy to digest, but in real life, there is a lot more emotional baggage to be carried. So first and foremost think long and hard about how much drawdown you can withstand.
What we have found that works with growing an account faster, is to divide capital into two pools, Initial Capital (IC) and Realised Profits (RP).
Step 1:
IC is the initial capital you decide to employ. With this capital, you will use a standard position sizing model, such as Fixed Fractional Fixed Percentage allocations.
Step 2:
Trade the strategy using the chosen position sizing model until you have a 10% realised profit. Assume $100,000 as IC which is now $110,000.
Step 3:
Move 50% of the profit into pool RP. IC is now $105,000 and will continue to be traded with your chosen position sizing method.
Step 4:
RP is now $5,000 and it's this pool that we'll be trading aggressively. Calculate the Kelly% (K%) across the last 100 trades. For this example we'll assume K% = 20%. In theory, and in the context of what K% was designed for, you would trade the RP with 20% risk per trade. However, in reality, we're going to trade 0.5*K%, or 10%.
At this point, assuming fixed fractional position sizing, the next trade risk will be:
IC = 0.02 * 105,000
RP = 0.10 * 5,000
Total risk = IC + RP = $2,600
Step 5:
Now we need to start re-balancing across both accounts. Therefore,
(a) When the balance of IC increases by another 10%, i.e.$115,500, move another 50% of those profits across to the RP account.
(b) When the RP account value is 50% of the IC account value move 50% back to the IC account and trade the new value with your chosen position sizing model.
If we continue to trade the RP account at the 0.5*K% level it will eventually make the IC account meaningless and the volatility starts growing exponentially. By redistributing the funds back we keep the broad volatility down, in other words, the aggressive RP account starts to feed the conservative IC account, and should a drawdown come along it won't set the total portfolio back too far.
Obviously, there are many ways to accommodate account growth but the bottom line is that more risk is required and ideally that risk should be calibrated to some extent so as not to blow the account out or cause irreversible psychological damage.
Oh, and I included historically listed stocks to avoid survivorship bias.So I thought I'd do a rough and ready analysis of a true random system. Here are the results....not very pretty is it? Think I'll pass on taking tips from a monkey throwing darts at a list of stocks or give serious consideration to a random trading approach. Unlike other so called random systems, my simulations were across all ASX equities (not just carefully selected constituents of a major index). My buys and sells were not carefully timed to occur at the start/end of a particular period (start/end of month). The system just made a random decision at the end of every day whether to buy or sell. To keep things simple I started with $100k capital and traded fixed $ amounts of $5000 for a maximum number of consecutive positions of 20. Trade commission was $15. I also used position size shrinking where I didn't have enough capital to take a position. I ran the simulations from 1/1/2000 through to 9/4/2021. Every single run outside of the 99 percentile lost money and that was for a 1000 runs--so 99% of the 1000 runs were losers. I'll pass on random systems.
View attachment 122607
MA,Oh, and I included historically listed stocks to avoid survivorship bias.
Month 8
Beginning of month 8. Like my weekly system, and CFD trading, I lost half of my open profits recently. I am liking this strategy though. I look forward to the EOM to adjust the portfolio and see how it is going. While technically doing worse than the XKO now, I think it will be a short setback. Still at approx 10% returns. Talking to friends, they find this return to be amazing. It s nice to talk to non-traders and have it in perspective that with just a few trades at the end of the month I can do better than most. I am confident that my system will end the year closer to 20% return. All speculation but I genuinely think it is doing well.
So I thought I'd do a rough and ready analysis of a true random system. Here are the results....not very pretty is it? Think I'll pass on taking tips from a monkey throwing darts at a list of stocks or give serious consideration to a random trading approach. Unlike other so called random systems, my simulations were across all ASX equities (not just carefully selected constituents of a major index). My buys and sells were not carefully timed to occur at the start/end of a particular period (start/end of month). The system just made a random decision at the end of every day whether to buy or sell. To keep things simple I started with $100k capital and traded fixed $ amounts of $5000 for a maximum number of consecutive positions of 20. Trade commission was $15. I also used position size shrinking where I didn't have enough capital to take a position. I ran the simulations from 1/1/2000 through to 9/4/2021. Every single run outside of the 99 percentile lost money and that was for a 1000 runs--so 99% of the 1000 runs were losers. I'll pass on random systems.
View attachment 122607
I think you're comments are highlighting exactly the issue I have. You are making refinements to what I believe is a better indication of a true random system to improve it by address key issues with randomness and those improvements are introducing an edge. For example you're suggesting moving to a weekly time frame to reduce trade frequency and then using XJO as a reference. I understand what you're saying and I'm not disputing that it doesn't make sense but to move beyond that and say a random system can beat the index, well I have my reservations. The simulation results I posted clearly show that in about only 1% of cases can a random system even generate 0.5% profit over 20 years. Now we can certainly start discussing characteristics of that random system such as trade frequency, time frames etc etc to see if we can improve some of it's undesirable characteristics but to me that starts to move away from a true random system. And yes it may well be possible to approach Index returns, but as I say all that is being done there is to address some of the negative impacts randomness has to boost returns. BTW, I think you review is correct and I'm not disagreeing with it.A few things to highlight
Your commission is 0.33%, currently IB are 0.088%. incl GST
Trading 20 positions, you have roughly 100,000 random selection events ( not trades just potential events). 20 per day *250 per year *21 years.
I don't know your sell condition but If your sell condition is "Sell = random()>.5;" you are turning over 50% of trades, that's 10 buys and 10 sells per day, and giving up 0.33% of your account each day to commissions. Not saying that's what you do
Turning to the actual market.
The XJO (asx200 index) made on average 3.62% PA over that 21 year period, that's an average growth of 0.014% per day.
To break even and overcome commission losses, you will need to limit trade frequency to 1/40 .... (1 trade in your 20 position portfolio per 2 days) your sell condition should be "Sell = random()>.975;"
If you change it to a weekly system, the average per week the index makes increases to .067% .... your trade frequency increases to 2.5/20 per week or "Sell = random()>0.9;"
This should get you to break even point using the XJO as a reference.
Now if you suddenly find that you are making money ( even 1% PA ), then the random market as a whole is beating the reference index. If you are losing money, then the random market is falling behind the reference index. ( If my maths is right ).
How to make money in the markets (trading or investing)
Watch YouTube video's & you'll be bombarded with "100%" foolproof methods that will make you extremely wealthy (for a price). I've only found one method (for me) that constantly generates a positive return over the long run, that being "system" trading.
System trading
Mechanical trading systems are systems that generate trade signals for a trader to take. They are called mechanical because a trader will take the trade regardless of what is happening in the markets. My all-time favourite is "Trend trading" which is simply trading with the majority.
Skate.
'an edge', 'random systems', changing nature of markets
The ratio of signals
When there are more buy signals than sell signals, which indicates "to me", the markets are on the improve.
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?