The Finance Book Club

Now we’re setting up for a bullish entry.
So much set-up. So little action.

Finally, (finally!) on Sept 12, 2007 a trade is entered long.

Entry statistics for this trade:


Afterwards, stops and alerts are adjusted daily. Price versus TS is monitored to make sure it is within limits.

By Sept 28, 2007 the trade enters ‘preserve mode’ and we get a ‘free trade’ meaning a profit is locked in even if the trade gets stopped out.
“Notice that tile closing price is 1.4267 and the value of the Tenkan Sen is 1.4053, which is a 2 14-pip difference. Our trading plan had stated that if price was 200 pips away from Tenkan Sen, we should exit. As a result, we should exit, right? Yes, because that is in our trading plan.”
And then…

“If you reached this point and have not noticed a mistake then you are not paying attention to your trading plan at all. I am purposely putting errors all over the place to make sure you follow your trading plan.”
Oh, this is grade-school BS! MP can take his patronizing attitude and stick it right back up his trading plan.

Here is the trap MP laid for the unwary reader:
“In our trading plan, it mentions that you exit when price is 200 pips or more away from the Tenkan Sen when it is flat. If you look at the charts, the Tenkan is not tlat so we should not exit.”

MP makes himself an unreliable narrator at this point. He can blame any problems with the book or the system on his clever ruse, and avoid taking responsibility for them.

I still want to take this outside the realm of the theoretical and see how his stated performance compares to the actual trading opportunities of the time, but for now I’m unhappy with this book, and this system seems both timorous and excessively management-heavy, especially from this algo trader who prefers to do as little busy work after the initial setup as possible.

This trade exits finally on October 3, 2007, with a profit as shown in 3.60



MP continues with more detailed descriptions of his 2 year backtest but this is exhaustive (and exhausting) and I think we understand as much as we are going to understand at this point. It’s nearly time to do some examinations of our own. First though, let’s see what he thinks of his 2 -year performance

I cant say anything about this coz i dont rely on MAs much
I dont use them for any kind of confirmation
Best thing about IKH is u dont try to sell or look for reversal as long laaging line green one is above the candles i was having some issues regarding reversal i think it will work lets see on live chart i will share my entries

Chapter 4 - Post-Analysis

1. Post-analysis is crucial as it gives the trader ‘the complete picture’
2. MP took a grand total of 8 (eight) trades in two years.
3. He was profitable so that is a good ‘first start’ he says
4. The problem is that there was too much risk vs. profits (!)
5. R:R was 1.51

It looks like there were two losses out of eight trades. One loss was significant, but smaller than the biggest win.

Based on his results, MP makes some adjustments.

“The goals of a trend system are to minimize losses when you are wrong and maximize profits when you are right.”

1. R:R needs to be < 1.00 or the risk is too great says MP
2. Winning percentage doesn’t matter says MP (true)
3. Looking at the data the first task is to minimize losses
   1. There were 2 losses in the original plan
   2. The bigger loss was 21 percent of the total profit
   3. MP says he must turn this big loss into a profit (!)
   4. The max profit for that trade was only 63 so this seems infeasible
   5. The second option is to eliminate the loss altogether
   6. MP figures the trade had an entry risk of 340 pips
   7. This is above the 300 pip value
   8. We allowed the trade because our trading plan did not include the buffer in the bullish and bearish rule.
   9. If we adjust our trading plan to include the entry buffer on the rule then we can eliminate the trade completely. And likely eliminate just as much profitable trading activity as well.

   
   

The philosophical problem I have with all of this is it assumes that some entries are of higher quality than others. Maybe on such long time frames and with such infrequent trading, that's possible. My guess is with more trading data we’ll find MP’s success ratio gravitates back to the 50-66% range, despite his ‘entry buffer’ tweaking. So is it really worthwhile missing all the trading opportunities that he does by trading this way? Some discussion of trade size would certainly be germane here, but I don’t see any mention of it in the table of contents, sadly.

MP continues to make revisions and decides that the May-September trade-less gap in history was a problem (yes) and so he decides to remove the rule that kept him out of trading during that time - the 300 pip max distance from the KS to the price.

MP then performs a new backtest, but on ‘stale’ data, which doesn’t tell us much. He needs to do a walk-forward test to have some sense of how the new rules will actually fare.

The new rules increase the number of trades taken, and doubles the anticipated profit, according to him. R:R goes below 1.0. He goes further and reduces his entry risk, changing max entry risk to 250. He also chooses to exit any trade after it is in profit by 550 pips.

MP acknowledges that he needs to backtest everything ‘again’ (but doesn’t mention fresh data) and even worse, he is going to leave that job up to us. We can verify the results on his website he says, but guess what.

The website redirects to some other outfit and mentions nothing about MP, or this backtesting. Oh well, I was planning on doing my own test anyway.

MP concludes the chapter by offering some more ideas to test:

1. Increasing the preserve value
2. Increasing the bullish or bearish exit buffers
3. Using the TS as a stop if price is 200 pips away from the KS
4. Using TS as a stop for continuation trades
5. Using the TS for all stops
6. Use the KS as a stop with no preserve mode

This mirrors my experience where we can think of things to try much more readily than we can find time and resources to test out our ideas.
MP then teases us with four other ‘optimization techniques’
1. Chart Patterns
2. Multi-Time Frame Analysis
3. Entry Strategies
4. Advanced Money Management (Scaling In and Out)

But as far as I can tell they aren’t examined in the rest of this book. Instead we are treated to more Ichimoku strategies (of which MP basically admits he’s already shown us the most reliable), something called ‘time elements’ (which MP admits he doesn’t use and prefers Gann timing) and of course, applied psychology, a chapter actually written by Doug Laughlin.

Frankly, none of that additional material/padding is very interesting to me until we can assess how well the ‘baseline strategy’ actually performs. So let’s attempt that, if possible, and then if it seems worthwhile, we can revisit the finer points.

It might take me some time to get around to this though as I'm focused on putting the finishing touches on a new algo (most promising so far, I hope) and then the holiday season beckons. So this might be a bit of a hiatus. Maybe OwaisJam will use my downtime to post some updates on his Ichimoku explorations.

If I don't get back around to these parts till after the new year I hope it's a good and profitable one for you.

Hallo @clemmo17, how do you do?

We await your return. Enjoyed your writings.

And we have seen it today, in fact, plunging all the way down to 1.09. Scary stuff for Euro holders!

Does anyone want me to revive this thread? Should I finish my testing of Ichimoku or just go through another book?

Could you give a list of books to read in currency trading for beginners?
And do the same for somebody at the intermediate level?
The better ones if you have.

The table of contents in the first post of this thread.
Start with the one's e has rated A.
I have read the Brent Penfold one. It has a lot of useful stuff

I read the comments clemmo made on Penfold book, it's definitely what I need.
But I wish there was a list made of books on currency trading only. One that would be divided on quality as well as skill-level.

The best books for beginners IMHO
Read #1 first then read 2 and 3 in any order.

1. Reminiscences of a Stock Operator by Edwin Lefevre
2. Universal Principles of Successful Trading by Brent Penfold
3. The Art of Currency Trading by Brent Donnelly which is specific to Forex but not retail forex. I have yet to read an outstanding book about retail forex trading, but I will be sure to tout it if I do.

Brent Donnelly is active on Twitter and I recommend signing up for all his newsletter material. His latest experiment is struggling out of the gate somewhat though.

Ok, I am going to try to finish what I started with Manesh Patel's system using Ichimoku.

Just to recap, all the rules we'll be following can be found in post #628.
https://www.forexfactory.com/thread/...7#post13800987

First, let's look at the period of history he himself tested using EURUSD on Jan 1, 2007. I am using the Ichimoku indicator that comes with MT4 but tuned to match Patel's settings.

Guide to the colours chosen using MP's recommendations.

If we examine the chart screenshot we can double-check MP's assessment of this position from post #629.

 

1. Price versus Kumo Cloud: Bullish (price is above the cloud which is yellow)
2. TS versus KS relationship: Bullish (the starting line obscures this but tenkan sen is above kijun sen)
3. Distance from TS: (actually my measurements are a little different, but this is probably broker data variance)
4. Distance from KS: (I am not prepared to sweat stuff this small with a system that is already ...detailed)
5. In Kumo Cloud: No
6. Tenkan Sen Trend: Yes (upward pointing)
7. Kijun Sen Trend: Yes (not flat since Dec. 22)
8. Chikou: Bullish (above price from 26 days ago)
9. Horizontal prediction: Weak, can run into price if it consolidates for four or more days. (more like 1 day to me?)
10. Vertical prediction: Weak, a 100 pip movement down can make the Chikou touch price from 26 days ago. (looks more like 200 pips to me??)
11. Trending Criteria requirement: (I guess?)
12. Future Cloud: Bullish. (yellow!)
    https://www.getyarn.io/yarn-clip/13c...8-0c078f7e29ab
13. Senkou A Trending: Yes: Pointing in direction of trend.
14. Senkou B Trending: Yes: Pointing in the direction of the trend.
15. Thickness of Future cloud: Thick: Indicates that the trend may not be a long-term trend, more medium- to short-term trend.
16. Kumo Shadows: No: within a 12-month period. (I guess??)
    
    Ok, good enough for me.

MP waits a whole 2 months before the portents align sufficiently to enter a trade. I've marked the day and the levels of entry/stop. This looks like a perfectly cromulent trade to me, except that the trend is fairly mature (up for 32 trading days since the January low) so I would expect some reversion.

He tightens his stop in April to the tenkan-sen, which on my chart looks like it would have stopped out a bit earlier, maybe Apr 26, but the level is basically the same. This is just a superior trade that anyone would love to have, nearly 350 pips.

MP takes 8 trades over 2 years and that looks like this. Template file attached.

Note the humongous period of inactivity between 2007.11.21 and 2008.02.27 and an even more titanic gap between 2008.04.24 and 2008.09.29. I could only trade this way if I was occupied with other instruments.

Over this period MP makes just over 1500 pips.

Now there are 2 things I want to test later this weekend.

1. Can we reproduce the amazingly good selectivity of MP's trades by using his rules and forward testing over a different period or does this performance require an amount of subjective 'expertise' that transcends just the system rules?

2. Can we obtain profits of a similar nature simply by following a set of simple rules and a favorite moving average (like the Hull MA)?

I'm going to start with #2 because it's easier to test.

First, I hide the ichimoku indicator. I'm still using Patel's 2 year history period from 2007.01.01 to 2008.12.31.

I replace the Ichimoku with an HMA using period 34, a completely arbitrary number that I know from experience is good enough for most currencies.


Here is my attitude to moving averages in general: they don't anticipate anything. Trends exist. They last a variable amount of time, and then they become ranges. MA's can only tell us whether we are currently in a trend or a range, not what's coming up. So, changing the period to a shorter value will filter out fewer counter-trends, and produce a choppier line, while a longer period will filter out more countertrend noise and produce a laggy line. You are choosing between getting chopped up in ranges or getting out of a trend too late to preserve all your profits. It's always a trade-off, mostly unsatisfying, which is why I rarely use MA's in my systems anymore. It doesn't matter which MA you use, they all operate on the same principle and have the same defects and benefits.

Having said that the HMA remains one of my favorites (probably #2) and it's common enough that you can find one easily on the internet. This particular version I'm using does have a bit of a twist which you can see in the settings panel. The 'speed' value lets you increase or decrease the amplitude of the line, giving you some interesting options when you have systems that exit when price touches the MA line. For this reason my HMA might look slightly different than yours with the same period and chart data.

However in this case my simple 'Patel Entire Ichimoku Replacement System Example' (PEIRSE) will not exit at contact with the MA line, but simply exits at a signal change.

Here are the rules which I think you'll agree are simpler than Mr. Patel's.

1. Use daily TF. Assume all trades are entered and exited at daily close.
2. Wait for a signal change (line changes colour)
   1. Note that like all indicators this one has a 1 bar lag between signal and price, so the bar that we consider for the signal change is the closing price of the bar after the division between colours.
      Attached Image

      
      
      I've tried to carefully observe this for all my trades but if I made a mistake please let me know.

3. To filter out choppy ranges somewhat, we require 3 consecutive higher highs for a long or 3 consecutive lower lows for a short trade. We count starting from the line colour division (so signal bar -1). In the example shown the signal bar is the second higher high and the following bar is the third, so entry would be at close of that bar.
   Attached Image

   

The first trade uses the existing signal after the start of the new year. I'm not sure if this is fair, so if you think I should exclude it we will. In any case it's only a small win.

Let's look at the second trade.


The signal flashes bullish at close of 2007.01.30 so the first trade exits.
The signal bar has a higher high than the one before it which has a higher high than the one before it. We need one more higher high to enter. The following bar (labeled #3) has a higher high so we enter at the close of that day or open of the following day, which is probably more realistic. Fortunately, daily gaps in price on currencies are very rare, and if we are trading the D1 timeframe, a day's price difference hardly matters when we are expecting to hold a trade for weeks and months.

Completing the 2 year period we end up with something like this.

A total of 19 trades, with six losses. The template is attached if you want to check my math.

You will note that this system has more trades, but also more losses than MP's system. It also has a higher loss percentage - of about 1/3 versus MP's loss ratio of 1/4. My losses are also larger, and occur in clusters whenever price ranges. A typical MA system result. So clearly MP's system is superior, right?

We should keep in mind that the point of trading is to produce profits, not just to avoid risk, although that is possibly just as important, if not most important.

If we look at the pip totals for this system and MP's we see that this system produced just under 2500 pips, versus 1500 for MP's system. So even though his system was able to avoid some losses, it also avoided some good trends. I should note that when I double-checked his trade results using the same rough close-of-day methodology that I'm using here I got something closer to 2000 pips for MP's Ichimoku system, not 1500 (which is his figure) so either he is losing profit by being picky about something or he had a data error (there are a few in his book) or I made an error (impossible!). Either way let's say that he had 2000 pips and I should reduce my count by 500, and you can see that we are either very close or slightly better trading this history period. And where his system is complex and effortful, mine can be easily automated. Should we? Chime in if you want to see an EA for something like this.

MP's results

My results. You can see the mean gain and loss are pretty close but the data skews toward a few big trends. This is the heart of trend-trading. You expect one or two trades to make up for a disappointing whole.

So far, in my opinion, all this supports my notion that being very selective about entry 'quality' is possibly counterproductive as it requires much effort and calculation (either human or algo) and in the end, we might just as easily miss opportunities as avoid crises. Personally I favour systems that remain agnostic to trend and simply use a statistical rather than a qualitative edge. Of course things might be different if we are trading in unusually turbulent times (like now) and using fundamental analysis, but that is not MP's style.

We should also keep in mind that 2007-2008 was a very trendy time for EURUSD. So those years naturally favour a system like mine, where a year like 2013 might be a miserable disaster for such a system if it didn't make adjustments. Maybe MP's system is better at keeping him out of unproductive dead zones, in which case it is worthwhile seeing if we can match his excellent results over a different data period (objective #1 from post #654).

The question is - what years should we check? We've seen what it can do during a tumultuous time, but what about if we use the relatively calm years of 2012-2013?

First things first. Let's check that his backtested entries conform to his own rules.

These are the rules, once again. I've numbered them so we can simply refer back to the numbers for each check.

Entry rules for a long trade -inverse for shorts

1. Price must be above kumo cloud
2. Tenkan san is > than kijun sen
3. Chikou is in open space (look back 26 bars)
4. At least 50 pips to the next ‘major’ S&R value (assume that's either, TS, KS or kumo?)
5. Kumo future is bullish
6. Entry has to be < 200 pips from the tenkan sen.
7. Entry has to be < 300 pips from the kijun sen.
8. Entry ‘buffers’ = 40 pips (MP never really explains this part, apart from saying it was determined by the ATR but I assume it means add 40 pips to entry and stop levels for additional safety)
9. All IKH indicators must be bullish (this is pretty much already covered in the above)
   
   Money Management
10. If price is > 200 pips from TS, exit if TS is flat
11. If profit is >= 300 pips, use TS with buffer as the stop
12. Exit buffer = 40 pips
13. Place or adjust the bullish alert
14. Adjust stops daily
15. Place alert at Kijun sen to cancel entry orders

This is the first trade entry:

Remember that we look backwards 26 bars to see the status of the Chikou - it is pointing into empty space above price, and we look 26 periods in the future to see the kumo status - it has turned bullish. So it seems everything checks out.

Trade 2:

Trade 3:
In fact it seems like he could have entered a bar or three earlier? Maybe the Chikou was still tangled up with price.