Spotting entry and exit points with heikin_ashi

[quote=“Falstaff, post:88, topic:122057, full:true”]

This is the huge problem I have had in the past in trying to do “trend following” - Getting stopped out and then watching the trend move on without me !

Have you done the exercise, if only in round terms as to the proportion of those “Stopouts” returned to the “trend” straight after stopping you out ? [/quote]

No, I haven’t done that exercise.

I regret to report that many of my positions are currently misbehaving. Realized P/L remains positive. Unrealized P/L has turned negative. Equity P/L (the total of realized and unrealized P/L, plus overnight interest debits and credits) has been positive right from the start of the test, and remains positive currently – but, just barely.

I have been taking positions based solely on the signals given by the weighted moving averages, without regard to trends or chart patterns. For now, I will continue to do it this way. But, I’m wondering whether other filters, such as trading only in the direction of the larger trend, etc., might be required to get the most out of this strategy. I will think about this, and address it at a later date.

[quote=“Falstaff, post:88, topic:122057, full:true”]

[Edit I think a catastrophic stoploss of whatever size is in order - simply so you don’t get killed by a “Black Swan” event, apart from that I agree “Let the trades breathe” and use your end of day decisions to govern the downside as well as the upside.][/quote]

Yesterday, I adjusted the stop-losses on all the open positions (except USD/CNH) to 300 pips (fixed, no trail). I did this on the basis of gut-feel, not on the basis of a rigorous historical analysis. That analysis is in the works, and may prompt me to change the stop-loss protocol once again.

USD/CNH and USD/MXN are special cases. The position in USD/CNH which is currently open has a 700-pip stop-loss. And USD/MXN (if, and when, I trade it) requires a 2,000-pip stop-loss.

Tripling the size of all the stop-losses obviously raises questions about overall portfolio risk, and I will be posting on that subject shortly.

Profit/Loss prior to opening new positions today:

Realized P/L: 69.5 pips profit

Unrealized P/L: -123.7 pips loss



Here is the entire watch-list, including open positions,
as of 5:30 pm NY time, Wednesday, December 6

  • EUR/USD - holding short
  • USD/JPY - signal to close long and open short
  • GBP/USD - holding short
  • AUD/USD - signal to close long and open short
  • USD/CAD - signal to close short and open long
  • USD/CHF - holding long
  • NZD/USD - range-bound; exit short; stand aside
  • AUD/CAD - holding short, despite loss and white candle
  • AUD/CHF - holding long
  • AUD/JPY - signal to close long and open short
  • AUD/NZD - signal to close long and open short
  • CAD/CHF - holding long
  • CAD/JPY - signal to close long and open short
  • CHF/JPY - holding short
  • EUR/AUD - holding short, despite loss and white candle
  • EUR/CAD - holding short
  • EUR/CHF - continue to stand aside
  • EUR/GBP - holding short
  • EUR/JPY - holding short
  • EUR/NZD - range-bound; continue to stand aside
  • GBP/AUD - continue to stand aside
  • GBP/CAD - holding short, despite loss and white candle
  • GBP/CHF - holding long, but chart looks toppy
  • GBP/JPY - signal to enter short
  • GBP/NZD - wait for new signal
  • NZD/CAD - signal to close short and open long; exit short and stand aside
  • NZD/CHF - continue to stand aside
  • NZD/JPY - wait for new signal
  • USD/MXN - wait for new signal
  • USD/CNH - holding long

Based on these signals, I have placed the following trades
(at 5:45 pm NY time, Wednesday, December 6)

  • USD/JPY (reversed: closed long and opened short)
  • AUD/USD (reversed: closed long and opened short)
  • USD/CAD (reversed: closed short and opened long)
  • NZD/USD (closed short)
  • AUD/JPY (reversed: closed long and opened short)
  • AUD/NZD (reversed: closed long and opened short)
  • CAD/JPY (reversed: closed long and opened short)
  • GBP/JPY (short)
  • NZD/CAD (closed short)

Currently there are 21 open positions.

Good to see you back Simon :sunglasses:

II’m sure @clint is trading this as a learning experience, rather than a cast in stone method and I’m content with adjustmets made “on the fly” when it appears something is wrong.

[Edit, myself I assess and place each stop as a trade specific assessment and will move them if I find something I had not considered. Stop placement is hugely important to me - second only to direction, and can and should allow for an entry based on an initial move against you ! - I have Never placed one as far away as 300 pips yet, but it is a possibility in the future, based on my statement below ! :wink:]

As a suggestion, I think the catastrophic stops should probably be “pair specific” - on the “Mex” 300 pips hardly seems adequate - whereas on EURUSD, 300 pips puts you either way over the 1.2000 swing high level, or way under the 1.156*** level (As I speak it sits a tad under 1.18000 ) - Then. whether it is suitable to include both of these pairs within the same “Test” is a question in itself - since a success or fail on the “Mex” could severely skew the result. - Just a thought. :slight_smile:

Notes on my test of the Captgrumpy Strategy.

In my description of this test, I have indicated that my watch-list includes 30 currency pairs, and that all positions taken are the same size. And I have reported the P/L (realized and unrealized) achieved in this test in terms of pips, but not dollars.

This test involves a portfolio of positions. Potentially, up to 30 trades (the entire watch-list) could be open at one time. This portfolio methodology is one that many readers of this thread may be unfamiliar with. So, I think it would be a good idea to detail (1) the metrics of the Oanda demo account I am using, (2) the metrics of the individual trades taken, and (3) the metrics of the overall portfolio.

The Oanda demo account came loaded with an initial balance of $100,000 (play-money dollars) – a ridiculously large balance for most retail traders to “play” with. I have chosen to trade this demo account as if it had an initial balance of $2,000. And I have decided that the position size for each trade will be 1,000 units of base currency. (Other brokers would refer to this size as a micro-lot. But, Oanda offers trading in units of currency, and doesn’t even use the lot or fractional-lot terminology.)

Here are the overall metrics for this test:

A maximum of 30 positions could be open simultaneously, with a combined notional value of roughly $30,000. This is 15 times the assumed $2,000 balance in the account, so that actual leverage used in this test could be as much as 15:1. Currently, with 21 positions open, totaling $21,588 in combined notional value, 10.8:1 actual leverage is being used.

Average used margin across the various pairs in the portfolio has tracked steadily at between 3% and 4%, corresponding to average maximum allowable leverage between 25:1 and 33:1. At present (Thursday, December 7), for example, the Oanda platform shows used margin as $812 – which is 3.8% of the $21,588 combined notional value of 21 positions. If 30 positions were open simultaneously, used margin would be approximately $1,200.

For each of the seven USD/major pairs, and each of the 21 major crosses in my watch-list, I am using a fixed 300-pip catastrophic stop-loss. For the two USD/minor pairs much wider stops are needed. For the USD/CNH (dollar/yuan), I am using a 2,000 pip stop-loss; and for the USD/MXN (dollar/peso), I am using a 6,000-pip stop-loss.

In a subsequent post (later this week, or this weekend), I will detail how I determined those catastrophic stop-loss levels.

It can’t be over-emphasized that catastrophic stop-losses are intended to rescue a portion of the account in the event of a black-swan event (such as the SNB debacle in January 2015) – not to provide protection against weekly or monthly price fluctuations, nor to function in any way as part of normal position closing. In this strategy, positions are closed (or reversed) based on once-daily, active trade management. Positions are not closed (or reversed) based on triggered stop-losses.

If all 30 potential positions were open at the same time, the implied risk appears, at first glance, to be $812.27 based on the fixed stop-losses listed above. See the table at the end of this post for details on how that figure was determined.

That implied risk assumes that all 30 open positions get stopped-out simultaneously. We know intuitively that such an occurrence is highly unlikely, and it turns out that it can be proven mathematically that the actual risk in a portfolio such as this one is far less than the implied risk calculated in the table.

For proof of this, I will refer you to a thread titled The Forex Portfolio - How to Gain Consistent Profits by Staying in the Market 24/7. That thread was started by Shawn Cannon almost 5 years ago, and it ran for about 3½ years. Shawn was active in this forum for many years, posting under the username mastergunner99.

The question of how to calculate risk in a portfolio of positions came up in mastergunner’s thread, and I decided to tackle that question back in 2013. The math got pretty hairy, but – if you want to wade through it – you can read about it HERE.

If 30 positions were to be stopped out simultaneously, based on the extraordinarily large stop-loss levels listed above, total losses in this test account would be $812.27 as noted above.

At the time of total portfolio stop-out, equity in the account would be $2,000 - $812.27 = $1,187.73 (ignoring all P/L booked to the account in the interim). Immediately after total portfolio stop-out, that equity figure would become the remaining account balance. There would be no open positions.

59% of the initial $2,000 balance in this test account would have survived.

At no point in this scenario would forced liquidation (a margin call) be a factor in the metrics of this account. Prior to stop-out, maintenance margin on the (hypothetical) 30 positions in this account would be 50% of initial required margin. That is, maintenance margin would be approximately $600. The lowest level of equity in the account prior to stop-out would be $1,187.73, well above the maintenance margin requirement.

Therefore, even in this absolute worst case scenario, no forced liquidation would occur.

Given these facts: (1) maximum, initial, required margin = $1,200 and (2) maximum real risk is well below the $812.27 implied risk (noted above), we can say that the $2,000 account balance assumed in this test is adequate to support this portfolio of trades.

From this we can extrapolate any number of account balance and position size combinations. Specifically, this test could be scaled up, or scaled down, to run in any size account, by adjusting position sizes accordingly. And, in an Oanda account, position sizes are infinitely adjustable. Here are a few examples:



How implied risk is determined.

Risk is defined here as the total of all the dollar-losses resulting from 30 stop-outs. Those dollar-losses are determined by (1) the pip-value of each pair, (2) the stop-loss for each pair, and (3) position sizes (1,000 units in each case).

The following table shows how these metrics combine to yield total implied risk – that is, the total loss which 30 stop-outs would produce, if such an event ever occurred.

Explanation of the figures in the table (using the yen-pairs as an example):

There are 7 yen-pairs in the portfolio, each having the form XXX/JPY. Any pair of the form XXX/JPY currently has a pip value of $0.0884 per pip per micro-lot (1,000 units).

$0.0884 per pip x 300 pips = $26.52 risk per yen-pair

$26.52 x 7 yen-pairs = $185.64 combined risk if all 7 yen-pairs hit their 300-pip stops simultaneously.

Later tonight, I will do the daily account review and update.

That review is going to be a little late this time.

Thursday, after the close, when I was unavailable to do the daily update, two positions were stopped out: GBP/CAD and GBP/JPY. This reduced the number of open positions to 19.

This update is being done at 2 am NY time, Friday morning, December 8.



Profit/Loss prior to opening new positions today:

Realized P/L: -611.7 pips loss

Unrealized P/L: 310.3 pips profit



Here is the entire watch-list, including open positions,
as of 2 am NY time, Friday, December 8

  • EUR/USD - holding short
  • USD/JPY - missed signal; close short for a loss
  • GBP/USD - signal (taking it late) to close short and open long
  • AUD/USD - holding short
  • USD/CAD - holding long
  • USD/CHF - holding long
  • NZD/USD - missed signal; wait for new signal
  • AUD/CAD - holding short
  • AUD/CHF - holding long
  • AUD/JPY - holding short
  • AUD/NZD - holding short
  • CAD/CHF - holding long
  • CAD/JPY - holding short
  • CHF/JPY - holding short
  • EUR/AUD - signal (taking it late) to close short and open long
  • EUR/CAD - signal (taking it late) to close short and open long
  • EUR/CHF - wait for new signal
  • EUR/GBP - holding short
  • EUR/JPY - signal (taking it late) to close short and open long
  • EUR/NZD - wait for new signal
  • GBP/AUD - signal (taking it late) to enter long
  • GBP/CAD - signal (taking it late) to enter long
  • GBP/CHF - holding long
  • GBP/JPY - signal (taking it late) to enter long
  • GBP/NZD - signal (taking it late) to enter long
  • NZD/CAD - signal (taking it late) to enter long
  • NZD/CHF - wait for new signal
  • NZD/JPY - wait for new signal
  • USD/MXN - wait for new signal
  • USD/CNH - holding long

Based on these signals, I have placed the following trades
(at 2:30 am NY time, Friday, December 8)

  • USD/JPY (closed short)
  • GBP/USD (reversed: closed short and opened long)
  • EUR/AUD (reversed: closed short and opened long)
  • EUR/CAD (reversed: closed short and opened long)
  • EUR/JPY (reversed: closed short and opened long)
  • GBP/AUD (long)
  • GBP/CAD (long)
  • GBP/JPY (long)
  • GBP/NZD (long)
  • NZD/CAD (long)

Currently there are 23 open positions.

1 Like

Profit/Loss as of the 5 pm close on Friday, December 8:

Realized P/L: -894.5 pips loss

Unrealized P/L: -393.5 pips loss



Here is the entire watch-list, including 23 open positions,
as of 5:20 pm NY time, Sunday, December 10

  • EUR/USD - holding short
  • USD/JPY - enter long on day-old signal
  • GBP/USD - holding long; ignoring reversal signal
  • AUD/USD - holding short
  • USD/CAD - holding long
  • USD/CHF - holding long
  • NZD/USD - enter short on day-old signal
  • AUD/CAD - holding short
  • AUD/CHF - signal to close long and open short
  • AUD/JPY - holding short
  • AUD/NZD - holding short
  • CAD/CHF - signal to close long and open short
  • CAD/JPY - holding short
  • CHF/JPY - signal to close short and open long
  • EUR/AUD - holding long
  • EUR/CAD - holding long
  • EUR/CHF - waiting for new signal
  • EUR/GBP - holding short
  • EUR/JPY - holding long
  • EUR/NZD - waiting for new signal
  • GBP/AUD - holding long
  • GBP/CAD - holding long
  • GBP/CHF - holding long
  • GBP/JPY - holding long
  • GBP/NZD - holding long
  • NZD/CAD - holding long
  • NZD/CHF - waiting for new signal
  • NZD/JPY - waiting for new signal
  • USD/MXN - waiting for new signal
  • USD/CNH - holding long

Based on these signals, I have placed the following trades
(at 5:25 pm NY time, Sunday, December 10)

  • USD/JPY (long)
  • NZD/USD (short)
  • AUD/CHF (reversed: closed long, opened short)
  • CAD/CHF (reversed: closed long, opened short)
  • CHF/JPY (reversed: closed short, opened long)

Currently there are 25 open positions.

I’m ending this test, and preparing to start over with a much simpler test.

I’ve come to the conclusion that it was a mistake to combine the portfolio methodology with a test of the Grumpy Strategy. Combining the two seemed like a good idea at the time, but I now think that true Grumpy Strategy test results are being obscured by the huge watch-list that I have tacked onto it.

Portfolio methodology is an excellent way to ramp up the performance of a proven trend-following strategy – but, as the past 2 weeks seem to have demonstrated, it doesn’t help in testing an unproven strategy.

Instead of making major modifications to the test that I have been running, I’m going to scrap that test, and start over. So, as of now, the previous test, with its 30-pair watchlist, is abandoned.

In a simpler, and more straightforward test, I will include only major pairs – and possibly not all of the majors (I haven’t decided about that). And I’m contemplating other changes, as well.

I’m just beginning to think about how to structure a new test, and I can’t say for sure when that test will be launched. I’m keenly aware that this is an especially difficult time of year to be testing new strategies. Some would even say that it’s an especially difficult time of year to be trading. If the Grumpy Strategy is as good as Captgrumpy believes it to be, then I’m sure it will have a better chance of showing its potential after the holiday season. But, I’m not committed to waiting that long. Everything is in flux, at this point, and the set-up for a new test is a work-in-progress.

I believe that several other people on this thread are experimenting with the Grumpy Strategy. I want to encourage them to share their results here. The more people involved in this testing, the better.

Hello Clint—Happy New Year to you!I have been watching with great interest as your testing of my plan unfolded…I am sorry to hear the results were not great,but I am not too surprised because of the large number of pairs used,some not of the ‘big eight’–The scope of your testing is far above my capabilities so I am very grateful for the tremendous work you put into it.I did a bit of testing on 4 pairs,but not happy with the end result of a small loss–I changed the WMA5 to WMA2 –
did some more testing with better results–The change to WMA2 shows a trade point one day sooner in many cases-(1 day chart)-this can mean a plus difference of up to 170 pips which helps a lot–It could help prevent so many stop loses you experienced.
-I have made another change adding the RSI4 indicator which shows a potential trade point up to 2 days before that
indicated by the WMA5–I currently have 3 active trades using the latest
RSI4 addition–2 trades are in the red -one with a small gain-I expect all 3 pairs to produce good gains now that things are back to normal.
I am looking forward to the new test you plan to do–with a smaller number
of pairs and of the top 7 or 8 currency pairs I think (and hope) the results
will be better–You mentioned one thing you may do is use a WMA7 in stead of the WMA5–might I suggest you look at using the WMA2–it shows the chart
reversal very close to the actual reversal time-I think the further out the WMA is the further it is away from the reversal-----As you stated others may be
trying my plan and it would be nice to see their results–I’m afraid I am not too optimistic that their results will be good ,but that is what is needed–if the plan is not working it goes in the scrap heap -just a few comments and a big thank you for all the work you have done!(and plan to do!)

[quote=“captgrumpy, post:102, topic:122057, full:true”]
Hello Clint—Happy New Year to you!I

Have been watching with great interest as your testing of my plan unfolded …

I am looking forward to the new test you plan to do–with a smaller number of pairs and of the top 7 or 8 currency pairs I think (and hope) the results will be better.[/quote]

Happy New Year, Grumpy!

I apologize for cluttering up my previous test with too large a watch-list. When I start a new test of your plan, it will be greatly simplified. In fact, I’m seriously considering limiting the test to just 3 or 4 major pairs. And I’m planning to stick to your original rules.

Nope. I think you have confused me with someone else. I have never considered the WMA7.

And I am not considering the WMA2, at this point. I think it’s important to thoroughly test one concept at a time. And, since this is forward testing, and not back-testing, it just takes time. There’s no way to speed up forward testing. We just have to wait for each day’s price action to play out. So, for however long it takes to determine whether the WMA5 (together with the WMA12) offers us an edge in the market, I will stick to the original rules, and resist the urge to tweak the test with the WMA2, the RSI, or any other “improvements”.

I’m not sure when I will begin a new test – definitely not before mid-January. Market conditions in late November through late January can be abnormal, not unlike conditions during the “summer doldrums” in June through August.

I’ll post here, giving you a heads-up, when I’m ready to start the new test.

Thanks for the update Clint–I understand your reasoning about sticking to my original plan–my only concern is that the WMA2 ,on my limited testing, produces far better results and could save a lot of time
Just as a matter of interest I am following 18 pairs using a greatly modified
system with the WMA2,WMA12,Heikin-Ashi candles,RSI4 combined with a
trend following plan.This is forward testing on paper trading --2 pairs started
the last couple of days in Nov–the rest all in Dec up to midnight last night (1 Jan).Most doing very well–three of these are active trades–2 doing poorly
because I read the trading signals wrong–they should start going the right way in the next few days–so far none have reached my SL set at 200.I’ll
list these pairs with results just for interest sake–maybe after you complete
your new testing on my original plan it might be worth while looking into
this new system I am working on.
AUD-JPY— +252 plus 97 (2nd trd) EUR-CAD— +182

  • 10CAD-CHF—+ 10 GBP-JPY— +289 plus 231 (2nd td)
    EUR-USD— n+269 GBP-USD— 112 plus 154 (2nd td)
    EUR-CHF— + 27 NZD-JPY— 362 plus 23 (2nd td)
    GBP-CAD— + 2 USD-JPY— + 78
    USD-CAD—+ 342 NZD-USD—+301
    AUD-CAD— + 41 ACTIVE TRADES --mini lots
    EUR-AUD— + 19 CAD-JPY— -102
    NZD-CAD— + 314 plus 92 (2nd trd) AUD-USD— -124
    AUD-CHF— -3
    Total pips gain on paper trades—( forward testing)—+3197 pips
    Total loss on active trades---------- -229 pips
    I will keep you posted on how the 3 active pairs do…

Hey guys…Captgrumpy, Clint.
I know it’s been a while, but, you guys been doing any work on this, this year?
I don’t know how I stumbled across this thread, but it didn’t take long to keep me interested. Because this is precisely how I trade. As I’ve read on and on, I kept thinking how both you guys have been going down the same road I have been, for some time now (most of all of 2017, up to the present). And here I thought I had the original idea, of trading the crossovers, along with the Heikin Ashi candles. Guess not.
Ok. Fine. :relaxed:
Anyway, this is my methodology. All you would have to do is check my long running thread to see when I started trading this way. But, seeing you two had such the interest (on much of my ideas), I just had to jump in here and hope to kick start this thread.
Look…if you or no one else is trading this way, or researching, backtesting, etc…it’s ok. Was just wondering.
This is how I approach the market. But, I do have to say, you’ve given me some things to think about.

Capt…
Clint…

By the way Clint, boy do I remember mastergunners’ thread. I was all over that!! And your input there was priceless!! I read that thread over and over again. But, as you have mentioned here, I agree that it’s not a good idea to combine them both. I think maybe that’s why I have picked one currency to concentrate on. (JPY)

Well, I would hope to see this thread revived again.
If not…it’s ok.
I just think I would be able to contribute much.

Mike

2 Likes

Hey Mike, it’s good to hear from you again.

Back in December, I told Grumpy that I was suspending my test of his method until after the holiday lull, and that I hoped to pick it up again in the new year.

Well, life intervened, and things got really hectic for me starting in early January. I’ve had to let some things go, and one of those things has been my intention to take a second look at the Grumpy Strategy (as I was calling it).

At this point, I don’t know when (or if) I might take another shot at it, but I’ll mull it over.

I plan to backtest this strategy with real tick data.
I coded the indicators, but not the strategy. I am still reading this thread.
@MikeWolski , what currency do you trade?

Hey Mihk
I trade the JPY.
Either long or short.
Thanks for the interest!

But I trade the daily time frames, only.

I think I got the HA Bar and LWMA for 1 Day timeframe calculated correctly. Switching my code from EUR to JPY pairs showed some annoying bugs I had to fix. I am still not done with the strategy. I will code it in the coming week.

I wanted however to post an intermediary backtest result I’ve done with the skeleton of a simple HA strategy. Enter/Exit at the change of the color of the Daily HA Bar.

gbp_jpy_2017 heikin ashi PnL 335.txt (6.9 KB)
usd_jpy_2017 heikin ashi PnL-749.txt (6.9 KB)
eur_jpy_2017 heikin ashi PnL-49.txt (6.9 KB)

I was surprised by the poor performance as I personally like the HA Candles. I think the main reason is that the price is already far away when you get a confirmation candle. Maybe this is also the reason captgrumpy strategy says to enter/exit when the LWMA5 makes a small spike up or down.

Hey everyone!

I have been reading through all the comments and your system has got me very interested, I will test and see for myself too.

I was wondering of any of you could give an update on how the trading has been, changes, etc?

Thanks for the good information.

Captain Grumpy may have been correct when he said the WMA2 was the better (faster) option to get in/out of trades.

I say this because

  • the HA candles alone are so far behind real candles in indicating change when used as part of a WMA system alone, and
  • up to 30% of pips in some trades are falling on the floor, because of these late entries/late exits, and
  • using WMA5 as opposed to WMA2.also means trades are missed or entered/closed a day too late.

The three bullet points above all point to the same problem: late entry and late exits have cost pips.

This is not the right place to say this, since it is a thread on spotting entries/exits using HA candles.
But since the thread has withered and has a very weak pulse, I’m going to be bold and say it anyway:
The strategy would be a killer if weak trends or whipsaws could be identified as an adjunct to the main theory.

As far as I can see, there is a lot to be gained here by the addition of a trend-strength indicator, such as the ADX.

I may begin a new thread incorporating the WMA2, instead of WMA5, and use the ADX, or Stochastic-RSI indicator as a filter, with respect to Captain Grumpy and all the above participants who so freely gave of their time here. Other than that, I propose to keep everything else intact.

I never did understand why the PTP was used - it seemed too circumstantial for me.
I would have thought that a change of candle colour would be a more verifiable concept.
We’ll see.

I may also include a comparison with the candle chart , as a filter to keep out of trades that have a high probability of whipsaw/failure.

I am sorry your efforts did not bear fruit - you certainly gave it your best.
Most of the proponents here simply burned out I believe.
Well done people. Thank you.

When/if I initiate a new thread, I will post a link to it here - I think this concept deserves further exploration.

Have commenced a Journal to discuss a similar approach combined with filters to ensure trend and try to avoid whiplash activity.

2 Likes