The "High Probability Entry" fallacy 

Paul King, April 28th 2008

A lot of time in trading is spent (I'd say wasted) looking for so-called high probability entries, but what exactly does that mean?  Generally it's trying to identify patterns in price/volume/fundamental/economic data that identify when there is a detectable bias to the future price trend of a tradable instrument.

There are a couple of problems with this approach:

1. Over what timeframe is this future directional bias supposed to operate?  If it's significantly less than your estimated average trade duration then what real use is it?

2. For every positive/reinforcing example of your high probability entry how many negative examples are there?  If this pattern only manifests itself in a small amount of data, or price moves counter to the predicted direction in about 50% of the cases, then is there really a cause/effect or predictive relationship going on at all?

If you do happen to identify a possible pattern with an edge, that doesn't have many counter-examples, occurs with sufficient frequency to test, and has accurate historical data available so you can backtest it, then try this simple test to see how effective it really is:

Once you've programmed your system/pattern and have a nice looking historical equity curve then, leaving everything else the same (i.e. instrument selection and ranking, position sizing, exit strategy), reverse the entry signal and see how this affects the results.

I've yet to find a long/short "high probability entry" that doesn't provide similar results when it's reversed leaving everything else the same.  What does this tell me?  Entry signal does not determine trade winning percentage or winner size - exits and position-sizing do that, so the term "high probability entry" is really an oxymoron to me.

What a particular pattern is really determining is a) the entry frequency, and b) whether you're getting into a tradable instrument that's moving about or stagnant.  Neither of these has anything to do with what happens next in any particular trade.  However, obviously a pattern that generates more entries in instruments that are moving right now has a better chance of capturing trends (in either direction) than one that generates less signals, or signals in instruments that are "stagnant" as far as price movement goes.

Stop "wasting" time on entry patterns and start working on position-sizing and exit strategies and you'll see an immediate positive effect on your trading.

Are you fit to trade? 

Paul King, March 31st 2008

Every time you sit down at your trading computer and make a trading decision you should have already asked yourself a simple question first:

Am I fit to trade today?

It's all too easy, once you have a well-defined trading method, or a few months (or even years) of experience behind you to get complacent and start thinking you're a mechanical trading machine that never makes any errors.

Forgetting that trading is a difficult activity to do accurately, consistently, and repeatedly without making any serious implementation errors can lose you many month's worth of profits in a single error.

When you wake up each morning take a moment while you're lying there waiting for your brain to boot up to think about how you feel.  Start off with a physical check:

• Does anything hurt or ache?
• Do you still feel tired?
• Do you have plenty of energy?

Then move onto a mental checkup:

• Is anything mentally bothering you?
• Are you ready and happy to go to the office?
• Do you feel alert and ready for the day?

I tend to simplify these questions into a daily "score" and then have some rules about what I can and can't do depending on how low (low is bad) my score is.  For example if I don't score high enough then I may decide to simply manage existing positions rather than putting on any new trades.  This means I'm not making important decisions when I'm not feeling 100%.

Obviously your fitness score can, and should, be changed as you go through your daily morning routine.  Sometimes, a little tiredness just needs a good hot shower to fix it.  Or some slow breathing when you get to the office may improve your mental state.

The important thing is not to be expecting full performance if you're not 100% fit.  It is OK to suspend trading if you're at a disadvantage mentally or physically that can more than eliminate any trading edge you have when you're 100% fit.  If you keep a diary of your "fitness score" versus any implementation errors you make then it should be relatively easy to see how your error rate increases as your fitness score decreases and then you can implement rules about how to reduce your trading activity when your fitness score is not high enough.

If you're trading when not fully fit you simply increase your chances of making an error, and that is the root cause of failure for traders that do have a complete and effective trading method already defined.  Don't take the risk, it's not worth it.

_______________________________________________________________________________________

Recently a trader I know made a small typographic error that cost him nearly $100,000 in 8 minutes and it could have been much worse.  Instead of buying a couple of hundred shares of something, he entered an order for over 100,000 shares.  This was because the order entry platform he used defaulted to a size of 100 and he didn't replace it with his required order size, but added his order size to the end by mistake.

A brief analysis of the 10 main errors that were made is summed up below:

1 He was tired, so shouldn't have been trading at all.

2 He was on the phone at the same time as entering orders, causing a distraction and a loss of concentration.

3 He didn't check the order confirmation immediately after he'd entered it.

4 It was right before the close, so it was too late to get out of the position when the error was detected.

5 He did not know how to participate in the end-of-day auction process to possibly exit the position immediately (this was not a US trade).

6 The market was very volatile (compounded by the large exit order) the next day so slippage was big.

7 He over-estimated the liquidity available at the open the next day and assumed that he would have similar slippage on exit to what had occurred on entry.

8 He entered a market order for the whole position instead of scaling out or using limit orders.

9 He did not know how to enter a pre-market auction order to exit the position with less slippage.

10 He didn't have a written plan for what to do in the event of this kind of simple trading error.

One mistake that he didn't make was that he exited the position as soon as he could.  Keeping the position (that was 50 times bigger than intended) to "see shat happens" could have turned into the kind of situation that puts you out of the trading game for a long time (if not permanently) rather than costing you a couple of month's profits.

The moral of this story is twofold:

1 Have a written contingency plan for every problem you can think of.

2 Have rigid procedures for why, how, and when you will put trades on and take them off (and follow them on every trade).

3 Analyze, record, and quantify your mistakes and put things in place to reduce the chances of them happening again.

In my experience most mistakes cost you money, are nearly always preventable, and if you learn from them, the chances of them happening again (or costing you as much) are significantly reduced.  Steps that should be put in place to prevent this kind of mistake again would be:

1 Have a method of evaluating whether you are "fit to trade" and if not, have someone else you can delegate the trading to, or even suspend trading for a while until you are fit.

2, 3, 4 Have a written checklist for trade entry (including confirmation that the correct trade was placed) and follow it on every trade.  Don't trade right before the close since you won't have time to fix any mistakes when they do happen.

5, 6, 7, 8 & 9 Always fully understand the dynamics, personality, and mechanics of your chosen markets before you start trading.

10 A full written business plan, operational plan, and contingency plan for your entire trading business is essential if you are going to avoid making mistakes that can easily overwhelm any positive expectation of profit from your trading programs.

Having a suite of trading programs that can make 20% return per year is useless if you make 30% worth of implementation errors!  Learning from, reducing, and attempting to eliminate implementation errors is a very important factor in success.  As Dan Harrington says about poker:

 "..in the very long run, your results at the poker table will approach the sum of your opponents' mistakes, less the sum of your mistakes".

You are never in control of your opponents' mistakes in trading, but you can try to ensure that the "sum of your mistakes" is a very small number and in doing that you will greatly increase your chances of success.

_______________________________________________________________________________________

It's hard to believe that it's been 7 months since I've had the time to post to my blog, but I've been a little busy developing a suite of trading programs for a US-based hedge fund I'm helping to launch for a UK client. If you want more details about that then please use the "Contact" page or shoot me an email.

For those of you that would like a nice summary of the previous blog entries I feel are most valuable and informative, I've published them in a little paperback called "Trading Truths" which contains all the best blog entries I've written since I started it in 2002.

if LowTrendLine > 0 then plot2(LowTrendLine,"Low Trend",DarkGray);
if LL2>0 and BarNumber=LL2Bar then plot3(LL2,"LL2",DarkGray);
            

_______________________________________________________________________________________

Consider the chart below (click it to download a larger pdf)

            

Consider the chart below (click it to download a larger pdf)

            

Evaluating the performance of your own (or somebody else's) trading is not as straightforward as it seems.  You need to take all the following things into consideration:

• The objectives of the trader or the method
• The risk adjusted return
• The margin usage or leverage
• The implementation errors
• The volatility of returns
• The time period of the sample performance

Unless you know what objectives the trading system or method was designed to meet, how can you tell if the achieved performance is good or not?  Achieving high returns with low drawdowns is not necessarily indicative of a "sound" trading method - it's likely to be due to luck, or trading methods where the small risk of a very large "blowup" just have not manifested themselves yet.

Return is only useful in the context of risk (e.g. drawdown).  Most sound trading methods have a return that is similar to the maximum normal drawdown.  Look at the largest drawdown ever in relation to the annual percentage return to give you an idea of risk-adjusted return.

Margin or leverage can be used to trade at a size that bears little relation to the actual amount of cash available.  This can distort any percentage return figures or drawdowns.  It's the ratio of the two that's important not the absolute percentages.

How many implementation errors are being made?  If you have a fully defined method it should be possible to determine how the actual return differs from the theoretical "perfect" return if you made zero implementation errors.  A sign of good trading is a gradual reduction in the number of implementation errors as the method matures and the trader becomes experienced at implementing it.

Volatility of returns is normally more important than the absolute percentage returns.  Achieving consisted returns is often better than bigger, but more volatile returns, to most traders and investors.

Any trading method will take a number of years to exhibit the true overall performance characteristics inherent in it.  Looking at small samples over only a few months is not representative of most trading methods, especially those that take a small hidden (but non-zero) chance of a huge loss.

There's more to trading results than compound annual growth rate, and one needs to do a detailed analysis of the trading behind the numbers to really draw a conclusion about what "good" looks like for any particular set of performance figures.

Trend following is a well known, widely used trading method, that goes into and out of favor like the wind.  Critics, academics, and traders alike either swear by it or discount it.  What is it about trend following that means it works at all?

Why trend following works is closely related to why it's psychologically hard to stick to.  With a low winning percent (usually around one third of trades) it means you get to be wrong a lot.  This may be demoralizing if you're more concerned with being right than making money.  Loser after loser can mount up and feel like the proverbial "death by a thousand cuts".  Months or even years can go by with nothing to show for your trading discipline except a smaller account balance.

All this for an infrequent huge boost from a small percentage of outsized winning trades that come along just after you've given up thinking you'll ever have a winner again.  But you kept trading anyway because you like the pain right?

This profile of lots of smallish losers interspersed with a few massive winners is exactly what makes trend following so hard, and also makes it a successful strategy.  To ride big winning trades you need wide stops, way outside the daily noise, and outside the inevitable two or three ATR reversals that most good trends contain.

Wide stops means smaller positions (if you risk a similar amount on each trade). This means you are less prone to a huge hit from prices gapping straight through your stop.  It also means trend following is pretty boring as trading goes, so if you're in it for the excitement you won't last long either.

All in all, trend following can be a very lucrative and successful strategy but only if you're psychologically adapted to the pain it can inflict on the inexperienced trader.  There's no gain without pain, and trend following can be financial agony if you're not mentally prepared for it.

In my previous blog entry I outlined briefly what pairs trading is.  In this blog entry I'm going to discuss some of the main assumptions that this trading method is based on so you can make a decision about whether it's a good idea or not for you.

Assumption 1

The correlation between the 2 instruments in a pair is permanent, stable, and long-term (at least for the life of your trade).

This is a big assumption.  Correlated price movement between 2 instruments is not predictable.  It could be due to random chance, it could be due to a fundamental reason that may change.  It could be a genuine correlation that will persist for the life of your pairs trade.  Beware choosing pairs that look like they are correlated but don't have a fundamental reason why they should be - it could just be random chance.

Assumption 2

Correlated pairs like the one described in assumption 1 exist.

If you cannot find liquid instruments that exhibit correlated behavior then this whole method cannot work.

Assumption 3

There are times when the spread between the 2 correlated pairs is "much larger than average" and represents a profit opportunity.

If the correlation between the pair is "too perfect" there will never be times when the spread is wide enough to give a decent profit by taking the trade.  This is a paradox of pairs trading - you need to find pairs that are closely correlated on a permanent basis, but not all the time!

Assumption 4

You are able to short the side of the pair that is going down.

Since a pairs trade is long one instrument and short the other, if you can't put on the short side of the trade then it's no use finding pairs.  This can be a particular problem for equity pairs trades where the short side of the trade is not available for borrowing from your broker, and does not have actively traded put options as an alternative.

Assumption 5

The pair does actually revert to the mean spread and you take your (small) profit. 

What if the "permanent" correlation between the two instruments has broken down and they will not "revert to the mean"?  You will have 2 positions that can both go against you.  Where do you admit that the trade is not working?  How much is the loss compared to the potential for profit?

Overall when one looks at the underlying assumptions of pairs trading it seems like a lot of things have to be true for it to work as a viable trading method.  Also with the small profit targets and tight stops the position-sizes tend to be very large in order to generate a decent return.  This can lead to a single large adverse event losing the equivalent of many, many, winning trades.

My advice would be to carefully evaluate the suitability of this method for your own trading purposes before risking any cash.

Assumptions are a big part of trading and understanding what assumptions a particular trading method is based on is a key part of trading success.  Some trading methods assume (and attempt to benefit from) trends or momentum of price movement while others assume some kind of "mean reversion".  Understanding which camp your trading methods fall into is an important concept.

Pairs trading falls into the "mean reversion" camp.  Pairs traders follow the method outlined below:

1 Find two instruments that have a close correlation of price movement (i.e. they generally move up and down together).

2 Determine what the variability of the "spread" between the prices of the two instruments has been historically by either dividing the price of one by the other, or deducting the price of one from the other.

3 Identify when the spread is above average (i.e. the prices have moved wider apart than normal).

4 Go short the instrument that has moved up and long the instrument that has moved down.

5 Capture the profit when prices "revert to the mean" i.e. the spread returns to "normal".

There are a whole bunch of assumptions that underlie the success of this trading method and in my next blog entry I'll discuss them and give you my opinion on whether they are useful assumptions or not.

1 - Unsound/incomplete method + Poor Discipline/Psychology = Failure

2 - Unsound method + Good Discipline = Failure

3 - Sound Method + Poor Discipline = Failure

4 - Sound Method + Good Discipline = Success

As we can see, without both a sound method and good discipline to implement it, trading cannot be successful.  Common reasons for an unsound method include:

• Incomplete trading rules
• Trading rules based on beliefs that are untrue or not useful
• Over-complicated trading rules
• Trading rules "curve-fit" to past data
• Trading rules that take too much risk per trade

Common reasons for little trading discipline include:

• Poorly defined objectives
• The wrong reasons for trading leading to self-sabotage
• Trading with "scared money"
• Plain old laziness
• Lack of education about markets, instruments, and trading in general
• Not taking responsibility for mistakes

Putting this all together to create a complete and sound trading plan that meets your objectives and suits your personality is not something that most people find easy to achieve without expert help.  This is the primary reason for trading failure in my opinion and why finding someone who can help you before risking real cash is usually a good idea (in hindsight of course).