The Once In a While Pivot Strategy.....

[GT3RS]

Hi all,

This is something I've been working on for a long time, and I finally found a way to make it work. The basic premise has been an idea I've had for ages, but I've nicked an entry method from a thread on [dare I say it!!??] Forex Factory!

It has been backtested on EURUSD only. From the start of this year to date, it has produced approx 250% increase on account balance, risking 5% per trade.

Drawdowns are low. Longest losing streak is 2 trades, so if you want, you could even risk more than 5%, but that's up to you.

This strategy doesn't trade very often, but is totally 100% mechanical, so lends itself perfectly to a Bot.

The stats are as follows [since Jan 1 2012]:

6.75 months worth of data.
18 trades
7 winners
9 losers
Longest losing streak = 2 trades
Risk per trade 5%
Account balance increase 250%
Risk:reward 1:6

OK. Here's the setup:

Timeframe:
M5

Indicators:
Market open lines
Pivot points - shifted
Stochastic set at 5,3,3
MACD_4CZ. Settings 5,34

Set the market opening times to suit your broker's server time [oh dear!! Here comes the first problem!! Oh no!! Sorry! I forgot! This isn't FF!!]

Set the pivots to print new levels at NY session close.

And you're good to go!!

The attached template and indicators will make your chart look like the one below.

Rules will be in the next post.

[GT3RS]

OK! So here's the setup!

Any questions, please ask!

Cheers,

Rob

[garyfritz]

Interesting! But 18 trades is a pretty small sample. I might try to automate this in Tradestation so I can backtest it on a longer period & more pairs. And so I can make sure it really is as mechanical as you say. (You'd be amazed how many times people say "it's 100% mechanical" but then when you test it, you find they subconsciously filtered out a bunch of losers...)

Questions:

"Price opens away from the central pivot point." Is there a minimum distance, or just "not touching" ?

"Price touches for the first time after the London open." How long before the London open does it have to not touch the pivot?

Other than that, looks pretty simple...

[GT3RS]

Hi Gary, yes! I agree, it IS a small sample, but encouraging enough to merit further investigation surely!!??

I'm hoping I haven't done any subconscious filters. I'm sure your questions and [hopefully!] the development of this thread will weed out if I have or haven't!!

As to your questions:

No minimum distance for the open price away from pivot. "Not touching" is a good definition.

Price can't touch the pivot AT ALL from the time the pivot prints [NY close] up until London open. If it does, there is no trade for that day.

Hope that clarifies!

[garyfritz]

OK, either I have a mistake in my code, or you've got some unconscious filters. I've got 32 trades YTD in EURUSD. It made just under 100 pips, but it lost steadily until mid-April, and won steadily since then.

BTW I closed trades at the NY close if they hadn't hit the TP or SL yet. Let me know if I shouldn't have done that.

Here's a trade listing. (Timestamps are GMT-6.) See how it compares to your test, and tell me if I have anything wrong. Once we're sure I have your logic implemented right, we can look at longer-term tests, other pairs, tweaking params, etc.

[GT3RS]

Thanks Gary. Wow! That's a big difference!! I'll double check everything when I get chance.

With regards to covering positions at NY close, I didn't do this. Once a trade is open, it either hits TP or SL. No management whatsoever once it's open.

Do you want to re-run your test before I cross check it against mine?

Thanks,

Rob

[garyfritz]

Hm, OK. Try this. It loses 15 of 16 trades from 1 Jan to 26 Apr, but it does great after that. Still 29 trades. Not closing at EOD kind of messed up my logic. I think I've got it right but definitely check the trades.

If I've got it right, you need to figure out what changed at 26 April. I see similar behavior if I look back to 1/1/10 -- steady losses except a streak in mid-2010 and a few huge wins in Sept 2011, then a bunch of wins after April. Overall it doesn't look very steady. You might get better results with a smaller TP and higher win% -- TP = 1 * SL looks better than 6 * SL. Setting the SL 8-10 pips past the swing also looks a bit better than 5.

Best case, it might make 500-1000 pips per year. Once we get this worked out, it may work well to run it on a collection of pairs. I tried a few others and it looked promising -- but not all pairs. E.g. USDJPY looked good, but EURAUD lost steadily. The optimal TP/SL settings might be very different to I showed above.

[GT3RS]

Hi Gary,

Many thanks for this.

I think you have uncovered the one area I have indeed got some subconscious discretion!! Whoops! Sorry!! You were right! Dagnabbit!! I was SO sure it was 100% mechanical!!

My [previously undisclosed!! ] rule is to not take a trade if price breaks the daily pivot by a "significant amount" before a signal being generated.

Hmmmmm...... so, what constitutes a significant amount? I have no idea, but I'll have a good think about it and get back to you. I'm thinking something in the region of 20 pips.

Please see attached chart. My broker time is GMT+2. Ignore the vertical lines saying Sydney Open and such, I only set the London one.

Would be interested to see your chart from the same day to see if piv is in same place and if all indicators are the same.

Cheers,

Rob

[GT3RS]

And here's my backtest spreadsheet.

[GT3RS]

Hi Gary! Me again!!

I've checked this out and I reckon we need to filter trades that set up after price has breached the pivot by 20 pips or more. I.E. if price breaks the piv by 20 pips or more, we don't take the deal.

This actually improves my results by taking out one deal that was a loser.

Also, no trades to be taken after London close.

And once you are in a trade, you leave it completely alone. It either hits TP or SL. No matter how long that takes.

Sorry for what appears to be additional rules, but they aren't additional..... I omitted to tell them in the first place...........

Are you able to code this in MQL4?

Cheers,

Rob