I have a suggestion for another (better) metric for trading systems. To me, the true measure of a system’s predictive power is what kind of market advantage it offers.
When I evaluate C2 systems - I generally ignore equity growth curves. The approach of some advisories is to:
- use a one or small lot approach
- compound the returns (can top a million dollars in less than a year if luck cuts their way)
- use all-or-nothing approach (like the famous “400,000 shares on one trade”)
Comparing ATDow or Consensus-Trading to some of these maximum growth seekers is like comparing a pea to a watermelon (not apples to oranges)!!
All I care about is, on a trade, is return on the underlying position. In other words, is it more impressive for a 6 hour ES long signal to go from 1200 to 1205.25 on 2 lots ($525), or to go from 1202.5 to 1203 on a 36-lot position ($900)?
Anyway, this suggested metric “equity curve” would ignore #lots, and only tally a return based on one lot (futures) or 100 shares (stock). Call it Best case, one-lot and put it next to Monte Carlo simulation (you can probably remove the New & (Whats This) to make room for another radio button.
The X axis would still contain dates, but the Y axis would show dollar growth based on one lot or 100 shares for each signal (most systems only trade one equity, but for those that trade several, you could also equalize different equities by adjusting for exchange minimum margin or stock price).
Yes, that’s a nice idea. I’ll try to add that very soon. (Just trying to finish the rewrite I promised my editor, first.)
If I may say. It won’t work for stocks. Most of stocks systems correlate position sizing on trading capital and volume of trading stock. 1 lot approach gives nothing in the case, imho.
For example, in comparing systems over a several month span, systems produced so many positive ES(emini) points on a one lot basis, and so many negative points. For several, I extrapolated from QQQQ or SPY to a rough emini equivalent. But this gave me a much better idea of how strong or weak a system’s predictive power is (it also makes it easy to calculate pure Profit Factor, which is a much better indicator of a system’s value than how much money an overleveraged system can rack up in a couple of months).
(other evaluated systems erased to keep this simple!)…
The point of this metric is to ignore position-sizing or other methods that mask the predictive power of the system. The existing metrics already handle position sizing. This new metric lets the reader see if a system has significant or weak abilities to extract trading advantages from the market (as my $525 vs. $900 example above).
Many of us probably ignore the developer’s . Frankly, I did some comparisons on systems that recommend larger or smaller sizes at times, and their larger positions did not produce a larger profit on a 1-lot basis - it might as well have been random.
For stocks, using a 100-share basis (or whatever method is used to balance across different share prices), this prevents the system developer from overleveraging or using methods.
For stocks or equities in general, another approach might even be to use a percentage basis.
For shorts use (entry price/exit price) and for longs use (exit price/entry price). This means all stocks or futures, regardless of share price or margin, are equivalent. Then the graph simply shows the percentage gains over time on a 1-lot or 100 share basis
But putting in anything for position sizing defeats the purpose of this metric.
One fly I see in the ointment is the size of the underlying security. For example, it would be difficult to compare 1 share of a penny stock .vs. 1 share of a $100 stock. In futures, corn wouldn’t compare to a full size S&P. Doing percentage gain comparisons won’t work either, as % changes in the Eurodollar (the interest rate future, not the currency) are extremely small, whereas % change in crude oil is large. Movements in oats are small enough that multiple contracts are needed for even small accounts to get any kind of meaningful return. I agree in principle with the suggestion, but I don’t think it’s possible to do like comparisons of different markets.
Simple example from my side. 1 lot of stock with price $100 - volume of 100K - price target 10% in middle of daily volume (there was 50K trading above your profit target). 2 lots of stock with price $50 - volume of 200K - price target 10% in middle of daily volume (there was 100K trading above your profit target). Both of the trades (it’s idealistic I can confirm it) have equal opportunity to have a fill.
Unfortunately, by your suggestion, first trade will be scaled on 100% and second one only 50%, but both of the trades are equal(!).
So what you’re going compare by the one lot approach? You cannot ignore volume in stocks it exists, pls believe me
yes - interest-rate futures (Eurodollars, TBonds) have a much slower movement. I tend to favor commodities in general, and I usually do separate analysis of interest-rate futures from the rest: grains, energies, softs, fibers, metals, etc.
Regarding penny stocks, the advisories on here seem to specialize in them separately from general stocks.
Granted, there may be a handful of advisories that may not benefit from this, but most advisories focus on trading 1 or several index futures or ETFs, where comparisons are rather easy.
To me, the bigger problem is the focus on pure dollar trading results, which tend to favor the overleveraged systems, which often actually have generally poor ability to extract a real advantage from the market.
In retrospect, I think that I might have misinterpreted the original suggestion. I took it as creating a master per unit number for the system as a whole, but I think you’re suggesting a column on the open/closed trade listing showing the per unit profit for each trade? If so, that would certainly make sense.
As Emily Lattela (aka Gilda Radner) would say: "Never mind…"
As far as futures trading goes, why not use a figure related to return on margin. For example, take a system trading the ES. The current system equity prior to the trade is $100,000 and the system goes long 10 contracts. The overnight margin is $3,150 (using SPAN) so the total margin is $31,500. This margin required is 31.5% of the total equity. Then if the trade makes a profit of $1000, the return on margin is 3.2% of the overnight margin of $31,500.
To me, those ratios representing margin to equity and P/L to margin are good indicators of risk and return that I can see at a glance. Using margin ratios can also help compare different instruments such as stock futures and bond futures.
Agree. Also, the margins are increased by the exchanges when necessary based on volatility (uncertainty) in the market, so the risk becomes self-adjusting and the return/risk ratio remains accurate.
Cheetah & Pal
Yes, that’s the best I have heard yet. The key for futures is to:
1) Make all types of futures comparable
2) Flatten out all futures advisories, so advisories compounding or leveraging to the max are directly comparable to those holding 1 position at a time.
And the corollary for stocks is: Profit divided by (share price X #shares).
So for 1000 shares of CompanyX starting at share price of $20, if the profit is $300, then return on equity is:
300/(20 X 1000) = 1.67%
This also allows advisories combining futures and stocks (penny or otherwise) to participate on a quasi-equal footing.
Matthew, are you getting this? The algorithm being worked out before your eyes : )
Matthew, are you getting this? The algorithm being worked out before your eyes : )
I’d prefer if the programming code were worked out before my eyes.
For futures, another idea, used in real time by regular brokerage firms is simply to double the margin requirements, thus not allowing over leverage.
Or also used by regular brokerage firms is to put a cap in margin % of the daily equity value. Many firms used 40 or 50 % on cash deposited in the futures account but also depending
of clients total net worth. This would be approximately similar to double margins ! And the first idea is more easy to implement.
To be even more specific, regular large brokerage firms such as Merrill Lynch / Prudential / Lehman/ etc etc have even 3 types of
margin depending of how much assets is either deposited at the firm or depending of how much the client can effectively prove as his real liquid net worth :
1) regular exchange margins
2) same + 50 %
And client who do not want to show anything of his real net worth but just want to deposit money and trade the futures market,
double margin are applied for initial/maintenance/day trade margin.
However this is never applied by online brokers.
The reason between the 2 types of brokers is simple :
- regular brokers will issue maintenance call when needed, but they will request the amount called to go back to the initial margins and NOT only to cover only the maintenance call. Really, so it is. Until a call is paid, the credit risk increases. I dont know what are the exchanges rules since online trades became very common, but for decades, the payment of the call up to the initial margin was an exchange rule, not a broker rule.
- online brokers are easier on their margin requirements, but they never issue a maintenance call, their computers liquidate straight away when needed.
Therefore, in my view, charging a minimum of 50 % additional margin makes sense.
On the other hand the weak part of this is that the cash equivalent of 1 NQ is half 1 ES but changes over a long period of time.
And 1 ER2 has a value of around 1 ES.
But the NQ and the ES have similar margin. But the worst is the ER that has the lowest margin of all, the biggest cash equivalent value and the highest volatility.
On that basis it seems to me that the fairest way to figure margin would be simply to say 10 % or so of the cash equivalent, thus limiting the over leverage.
The same could be applied for Forex, of course.
And on the stock side, one simple rule could be not to have more than X % of the account value in any one stock at time of the initial position.
But measuring things per 1 lot would not allow a money management, in my opinion.
your idea is why most people lose monet on trading. trading is way more then simple single dimention buy sell signals.
sizing is a crucial part of successfull system.
most of my trades are losses but from time to time we get the mega trade. that pay for everything and makes lots of money for me.
Why do you all want to invent the wheel? There is a common standard metric that solves all the problems you stated.
Profit or value / max DrawDown.
If it is over leveraged it will have big drawdowns.
It is also looks very small modification.
A very necessary addition is exponential charts. Non exponential charts are not giving the real picture.
Agree. A good position sizing strategy will result in greater, more consistent profits on a high expectancy strategy than on a low expectancy strategy, even if the low-expectancy strategy has a higher net profit on a 1 contract/lot basis.
"your idea is why most people lose monet on trading."
Actually, aside the fact most people are under prepared,
they lose money because they are under funded and over
leveraged. That is the fact and the brokerage companies
have done studies to prove this. So for the little guy looking
for a system seeing the one lot performance IS important.
Indeed, even with a $100,000 paper trading account some
of the system guys here went to nearly negative $100,000
(i.e. they had a DD of $200,000 on $100,000 in equity).
Show me a real life broker that will let you that ;-).
I agree with Jay. This whole discussion ignores 2 of the main aspects of a trading system: position-sizing, and risk per trade (in relation to reward and account size).
Any metric that simply uses round lots or single contracts does not take into account how effective the positon-sizing is, and how much risk per unit is being taken (based on initial stops).
Somone who is making 1 point on one S&P contract with a stop 10 points away is obvioulsy inferior to somone who is making 1 point with 1 point stops. This suggested metric completely ignores this (and renders it useless in my opinion).
I don’t want to start a big discussion/argument, just presenting my point-of-view. Matthew can decide for himself whether he believes this functionality will be useful or not.
PMKing Trading LLC