Well I’ve got a lot more to say about this but for now just one thing to note:
Remember that ALL of Wilder’s trading systems and indicators were designed to be used on the daily timeframes. And all of his indicators (ADX, RSI, as but two examples) have a default value of 14 DAYS. And that’s not a ‘thumb suck’ default. It represents ONE HALF of the trading period i.e. ONE HALF of ONE MONTH (on average). The figures vary but mostly you’ll find even with other trading systems (Larry Williams’ stuff for example): the default EMA periods that he uses now and then are either 13 DAYS or (more recently) 26 DAYS. Bill Williams: pretty much the same ‘deal’. What I’m saying or asking is that have you tried to see on the DAILY charts what ADX(14) is telling you and then only taking trades on the 30-minute timeframe in the direction indicated by +DI and -DI (and if you’ve got a LOT of patience only take these trades when ADX and / or ADXR is above 20 - 25 and rising??? Just a thought.
Let me say this though: ADX is by no means any ‘magic bullet’. It does lag. HOWEVER: as I’ve noted somewhere else on this thread beware of the ‘every last pip syndrome’. In other words: the ‘ideal of ideal’ situations is to take a trade when the distance between the +DI and -DI lines is increasing and accelerating and ADX is rising and accelerating (after crossing over from being below both the +DI and the -DI lines to above both the +DI and -DI lines) and is above 20 - 25. There’s no doubt about that. The problem is that it severely limits the number of trades or trade signals that you’ll get in any given month AND when looking at a chart you may want to ‘spit blood’ when you can clearly see that had you NOT been following ADX’s signals or using it as a filter you’d have missed the very beginning of a very large move. Personally: this is a ‘demon’ I’ve never been able to conquer. That being said: I’m really of the opinion that if you DO have the ability to take fewer trades, using ADX as a filter (with all of the above conditions being met), then you’ll in all probability find that your win/loss ratio would ‘skyrocket’ (in favour of the wins). At some point I even coded an oscillator for ADX that basically gave you a ‘go’ if the above conditions were being met and a ‘get out or at least move your stop’ signal the moment those conditions had changed (this on the daily timeframes and using the 14 DAY default) but with one look at this oscillator it was VERY clear how much ‘patience and restraint’ would be required to trade using ADX (and / or ADXR) as a filter (for ANY trading system). I just ‘don’t have what it takes’ to trade like that. It’s a bit weird really. And it REALLY comes down to this thing of ‘trader psychology’. Do you enter a trade KNOWING that you’ve got a high probability of a) getting into a trade far earlier AND b) that you’re going to get whipsawed!!! There’s a ‘conundrum’ for you!!! LOL!!!
But that’s why I’m now ‘banging on’ about Wilder’s book ‘The Adam Theory of Markets’. NOT so much for ‘Adam’ but for all the ‘words of wisdom’ in the book. Wilder says it OUTRIGHT that (for example) ‘Adam’ WILL be wrong at ALL market tops and bottoms BUT it will catch most of the move in the middle which is where the ‘meat’ is. And I guess the same logic could be applied to ADX really. Also: I REALLY DO recommend that anybody that uses any of Wilder’s indicators actually LEARN how they are SUPPOSED to be used and you can only do this by reading ‘New Concepts In Technical Trading Systems’. ADX is a very complex indicator and has many ‘nuances’ but for the most part it’s portrayed as a simple ‘go long if +DI crosses over and above -DI and ADX is rising’ (and again visa versa for short). There’s a lot more that ADX can tell you than that (which you may or may not want to actually see). RSI is another indicator that is largely misunderstood and if used CORRECTLY could result in catching some major moves. For one thing: this ‘thing’ of going long when RSI crosses the 50 line from being below it (and visa versa for short) is not even MENTIONED by Wilder. Somebody at some point in time just HAPPENED to notice that there MAY be some merit in this (but to the best of my belief soon ended up losing given that this was the core of their newly designed trading system and here I’m not referring to ONE person I’m referring to probably hundreds if not thousands of people).
And ‘tweaking’ the ADX period for shorter timeframes??? Well there’s a ‘million dollar question’. Could it / would it work if, for example, you said that your trading period was 24-hours long so therefore you’d use an ADX period of 12 (and assuming you’d be trading the hourly timeframe)??? I’m not convinced. As I keep saying: Wilder traded only commodities and only daily and ALL of his systems and indicators are heavily weighted in favour of the DAILY CLOSE. This has always been a contentious issue with me and everyone else but I do believe ‘it is what it is’ and until somebody ‘technically or scientifically’ proves to me that there is no merit in this and that the DAILY CLOSE is NOT more significant than an hourly close or a four hour close then ‘that’s my story and I’m sticking to it’. LOL!!!
By the way (given that I’ve now mentioned ‘Adam’ here again): I ‘solved’ ‘Adam’ on Sunday afternoon, for the week ending 09/12/2011, for EUR/USD and simply went short at market when I woke up on Monday morning and placed my stop where ‘Adam’ would have me place it. Lo and behold: that trade (as I type this) is worth +337 pips. ‘Beginner Adam Theory Luck’??? Time will tell I guess. And just to ENSURE that I cannot post an example of the trade when it is over and be accused of ‘curve fitting’ the trade ‘after the fact’ I purposely emailed someone with my ‘projection’ on Sunday. Alright: it’s just on a small ‘no deposit $20 bonus account’ (FOREX only unfortunately) that I found so I really didn’t care less what happened to it but thought I may as well use it to test ‘Adam’ (because it was useless to me for anything else). I ASSUMED that at this particular broker that a 0.01 lot size of EUR/USD was a nano-lot but not so i.e. they’re micro-lots so right now that account has more than doubled so I’m figuring that if they’re monitoring this little account they’re MIGHTILY ‘pis*ed off’ by now!!! LOL!!! To add insult to injury I should try to withdraw the $30 I’ve made in profit SO FAR and THEN see what happens (if I ever get it)!!! LOL!!! Alright: NO thought was given to risk or money management (well that’s not STRICTLY true i.e. if they WERE nano-lots then there would have been no problem but, and had ‘Adam’ been wrong, well the account would have gotten margin called WAY before my stop loss was hit so don’t go ‘salivating’ over the more than 100% gain in three days i.e. had they indeed been nano-lots then the account would be nowhere NEAR doubled)!!! LOL!!! But I’ll keep testing. Put it this way: let’s just SAY that ‘Adam’ has merit. Since Monday morning, trading EUR/USD, on ANY timeframe, I’d have only taken short trades (using this system that is being detailed here or, for that matter, simply using Parabolic SAR on its own). I mean: EUR/USD has gone down ‘three days straight’ so far which is exactly what ‘Adam’ predicted would happen (although I have to admit not to the same DEGREE as it HAS INDEED fallen). The point is: ‘Adam’ correctly predicted the ‘most PROBABLE direction’ that this pair would move in for the coming week. If there’s any merit to it: that’s ‘Gold’ man. That’s ‘Gold’!!! LOL!!!
Hi
I trade daily the dax,dow and eur/usd
I have found that by adapting psar from standard deviation to 0.10 AND 0.11 you get better results on
15 minute chart.
I tried useing 5 minutes but had a number of false trades.I now check 1m ,5m and 15min timeframes.
I use intertrader advanced charts and am only looking for 40/50 pips a day.
I use the psar as my stoploss and use a trailing stop set at 10 pips to try and maximise.
very good thread
Hi - I know this thread is basically dead but this strategy has some potential. All it needs is a few filters. - And trailing stops for taking advantage of getting more out of trend moves instead of keeping a cold SL/TP of 2/1
Here are some i’ve come up with:
Trade the USD/JPY 1hr only due to the fact that it is a trendy pair.
Trade the PSAR on a higher setting, a lot of entries with the default settings create losses wayyy too quickly and fail to capture trend impulses. I found that settings of (0.0025,0.2) work nicely but I would like to test others.
Go long when the PSAR reverses from above to below and the candle with the first PSAR dot below (entry candle) must be above the 240 LWMA (10 Day) -------- and vice versa
*Stops are at the PSAR dot.
Take 50% of your profit when price has moved 50% of your stop in your favor.
Take the rest when there is a close below/above the 50 EMA. (Disregard this until the 50% TP has been reached.) - Alternatively you could wait for the PSAR to switch.
I have a few more filters (like the minimum slope of the 240 LWMA, the maximum difference between the close of the entry candle and the 240 LWMA to prevent mean extension, etc…) but first I would like to make an EA for what i’ve described so far and backtest that to see if more filters are neccessary. Im going to try to learn MQL4 and will post results asap. I will also try to post some images to make the system more clear. BTW if u know MQL4 feel free to code an EA
Greetings Dale (LOL!!!). It’s ages since we last met at your place.
This thread has been dormant for years but I thought I might be able to add a new insight, since all of the above discussion is based on single currency pairs, whereas I am looking at a full basket of pairs treated as a unit.
I recently installed a PSAR EA on a demo account. It’s a plain and simple EA, no fancy filters, but it’s robust and does everything well. It has a PSAR trailing stop. I set it loose on 28 pairs, trading only the London and New York sessions on the four hour time-frames. That typically gave me up to 10 simultaneous open trades at a time. I noticed that on most days the TOTAL floating balance was almost always positive initially, even as no pair on its own was doing anything special. I soon realised that treating the entire basket of 28 pairs as a single entity and ignoring any particular pair seemed effective. Of the 5 days I’ve been playing with this (and all during this very slow end-of-year period) I have had 3 days >10%, 1 ~B/E (breakeven) and 1 losing day.
So it occurred to me that if I could have a setting in the EA whereby all trades for that day are closed when price reaches -x% or +y% (one would soon learn what numbers work during which conditions, I’m toying with -3% and +10%), then one might have a very useful tool. A ‘move to B/E’ setting after total profit moves to, say, +4% might also be helpful. At present I have been doing this manually.
I think it works because even on a short, counter-trend trade that would normally be stopped out (the bane of PSAR in non-trending markets) there are usually 3 or 4 PSAR dots initially before it reverses and gets stopped out. If one leaves such losing trades to ‘mature’ they will be stopped out, but for a while they hover around B/E and don’t represent a major loss. At the same time, there will be some other trades in the basket that are moving into profit. This allows an initial profit spike across the entire basket of trades that can be banked after 6 or 7 hours (that’s the theory, anyway!!). This is very early days and during an unrepresentative period of the year, but I plan to continue doing this to see what happens.
So, following on my previous post I have set up 3 demo accounts with the PSAR EA identical apart from the following:
No trend filter, so will always be ‘in’, manual closing at -3% or +10% (these numbers are provisional at present)
D1 PSAR trend filter (it’s provided within the EA settings) as a way of trading with the dominant trend, manual closing as before
D1 PSAR trend filter, no manual intervention leaving it to run, but within the same set hours as the others.
Based on consensus that an unfiltered, unmanaged PSAR strategy is not successful, I am not bothering to run this obvious experimental control. I should also have another one running on a 24 hour basis, but any more instances of MT4 on my machine (I have 7 currently) might cause it to catch fire!
I will be recording daily results on a spreadsheet and will report back perhaps weekly to let you know how it goes (if there is anyone out there still reading this old thread!).
For the record, here are my settings:
Trading TF is H4, times to place trades: 08h15 GMT to 15h00 GMT (all open trades will continue through the NY session until they hit TP or SL or I close them), lot size is 3% of balance, stop of 40 pips, TP of 60 for now, might increase or dispense with this later - remember I will be closing all trades at the end of each day. I’m not going for big individual trades, but am interested in what the entire basket of 28 trades does, so it’s going to be unusual (I think) for any trade even to reach 60 pips, (but if 20 reach 20 pips each …), PSAR is set on 0.035 and 0.8, rather than the usual 0.02 and 0.2, based on wanting to keep the dots closer to price movement to lock in profit quicker, D1 PSAR filter is set on default in order to keep a more continuous trend profile, move to breakeven is on 35 pips. Over time I have become comfortable with this fairly tight B/E setting - it’s always a case of compromising between giving a trade space and keeping things tight. I’ve been in a ‘tight’ phase over the last two years.
I think that’s it. I’m excited to see what happens starting tomorrow (Tuesday 2 Jan), even though the year doesn’t really get going until next week - can’t sit on my hands any longer!
Hi Alistair, I run a PSAR cBot (cTrader) with a very similar strategy AF settings @ 0.009 and 0.5 for one strategy and a “scalping” version in the tick charts with AF’s of 0.11 and 0.1.
Both use a 200 WMA Slope (Colour) to give trend guidance to the logic…ie: only Buy (PSAR Signal) on the MA Green and only Sell (PSAR Signal) on the MA Red… stops a lot of the false signals that can be generated by the PSAR when price is ranging.
I’ve been running a PSAR EA for the 4 days of last week and all of this week, on 4 different demo accounts to be able to compare different strategies.
The common features for all 4 are: set to place trades from 08h15 GMT to 15h00 GMT, set to use H4 for entries, risk set on 3%, PSAR trailing stop, PSAR set on 0.035 and 0.08, move to B/E after 35 pips. I focus only on the total profit, not individual trades, so when I judge the profit is decent enough I close all trades and leave that account blank until the next day.
The strategies are:
PSAR as is, no filter, but closing manually, usually around 15h00 GMT
PSAR with D1 trend filter, closing manually as above
PSAR as is, no filter, no manual closing, running full-time apart from the time restriction for placing trades
PSAR with D1 trend filter, running continuously as in #3.
The first 2 strategies started last week well, with #1 doing best. It made about 28% return last week, but lost small amounts on all 5 days of this week. It closed about 10% up overall, courtesy of the 28% it made the previous week. Strategy 2 made 8.1% last week, but then lost on 4 days this week, making up a little today. It ended on 6.8% overall (ending balance = $1068).
Strategies 3 and 4 were slow starters, but over this week have managed to build their equity amazingly. #3 ended the testing period with 80% profit (it’s true!), or a balance of $1802 made from $1000 since last Tuesday morning. Strategy 4 closed with a ‘mere’ 51.9% profit.
It is clear that the D1 trend filter reduces profits, but it might enhance reliability in the long-term - time will tell. The 80% return made by Strategy 3 is unbelievable, and it might prove to be just that - win big this week, lose big next week. I wish I’d had real money on that one, but next week I might be singing a different tune.
I recorded total profit for #1 and #2 each hour of the trading period. There seems to be an early peak between 10h00 and 11h00 GMT and a bigger one (which is when I closed all trades most days) between 14h00 and 15h00 GMT.
Hi Alistair, the PSAR used for position entry and exit definitely works. I had a poster a few months back trying to tell me it was only for setting Stop Losses…
My EA has a trade limit (robot stops after say 3-5 trades) and a position limit switch that sets how many positions are open at one time (default 1) with logic controlling drawdown. It also utilises Equity protection so if drawdown exceeds say 0.7% it closes the position and stalls the bot. Another switch stops the robot at the end of a trend if required, which will be useful on longer TF’s. I have it running on a few different TF’s with more success on 15min +.
The Slope MA controls the PSAR by stopping the bot trading against the longer trend.
Once again with this strategy I do not use TP or SL’s. The MA change of direction (colour) and confirmation by the PSAR will close or open the position. Chart below is the EURUSD (1H) prior to the weekend.
It is not always as pretty as above and It does get tripped up by price action hence I limit the amount of total trades depending on pair volatility. This strategy is still running on a demo and has added 23% to a $1000 account (0.05 lots) over 2 weeks, although it is not active full time.
Still requires a lot of logic coding to sort the issue when price is ranging. I’m trying to get the bot to understand the length and slope angle of a trend to filter out trading during congestion.
How has your PSAR bot being going? Would you be willing to send it to me? It has some functions that are not provided with the PSAR EA I am currently using and I’d be interested in using both side-by-side to see what effect the additional functions have.
