Tuesday, 29 November 2011

Buy signal at 17:45, filled at 2219, market immediately sold off taking me out in a matter of seconds, for a 3 point loss.

Further buy signal 17:50, filled at 2215.75, again with this one the market sold off taking me out for a 3.25 point loss.

A 6.75 point loss today (after commissions)

There was a further buy signal after this that also would have been stopped out, - not traded, thankfully.

Monday, 28 November 2011

I found an interesting website called www.capitalcontext.com, they have a model named CONTEXT that correlates with equities and yesterday their model diverged away from equities in a similar way to my indicator.

Someone asked them how their model is calculated;

"Consider CONTEXT as being driven by a set of non-linearly dependent (i.e. not linearly correlated) factors including FX Carry, credit, commodities, rates/curves, and swap spreads. Our framework learns individual relationships among all of these factors and ES in order that we can create a quick-and-dirty perspective on how a broad basket of global risk-drivers is behaving relative to US equities.
The chart (difference curve as you describe it) is always self-consistent – i.e. there is no running recalibration within the chart (or over-fitting in real-time) as, just as you mention, this would make the framework very hard to use practically (which is our simple goal for our own trading and hopefully for yours). Instead the effort is made on creating a process that uses our experience with these relationships and the market's empirical behavior, along with an understanding on when these relationships change regime, in order to know when a re-calibration is necessary.
The key for us is a framework that has an end-result that is practically useful (identifying market disruptions and pin-pointing specific risk drivers), actionable (whether for flow comprehension, risk support, or in our institutional client's case convergence execution), and easy-to-use.
We hope this helped.

They have a different set variables going into theirs however commodities might have been the cause of yesterday's divergence for both since Crude oil wasn't correlated to equities at all yesterday;

Further sell signal at 16:39, filled at 2224.50, market consolidated for a few minutes before selling off down to 2216.25, moved stop to 2224.25, market rallied back up taking out my stop for a 0.25 gain.

7 point loss today (after commissions)
Not a good day today.

Sell signal at 14:38, filled at 2207.75, market rallied sharply taking out my stop at 2211 for a 3.25 point loss.

Further sell signal at 14:48, filled at 2213.25, again the market rallied sharply taking out my stop, filled at 2216.75 for a 3.5 point loss.

There were two other sell signals, one at 14:56 and 15:24 however since the move up was so bullish I passed on those, both would have been stopped out also.

Quote from MarketWatch.com - "Stocks swept up in optimism wave", I've been swept away by that so far today.

Wednesday, 23 November 2011


Net gain 15.5 points, an ok start

The breakeven rule looks good, getting me out of the trades that aren't working

72.7% of the trades moved 4 points in my favour from outset (similar to the 72% figure achieved during backtesting)

Indicator has come up with some precise entry points
Sell signal at 18:03, filled at 2182.50, over the next few minutes the market went sideways, since it wasn't acting as expected I exited at 18:08, for a 0.25 point loss.  Just as well since the market then headed lower before trading up to the exact level my stop had been at, then immediately sold off down to as low as 2168.50.

 I won't be trading over the next couple of days due to Thanksgiving.

Tuesday, 22 November 2011

Sell signal at 18:25, filled at 2220.25, this was prior to the release of the Fed minutes,

"While  the  FOMC Minutes have not yet been officially released by the     Fed,  it  appears someone has broken the embargo. Here are the  headlines.

In other news - nothing substantially different from the statement or the conference that followed."

The  market was generally directionless, but it did sell off down to  2214    by 19:05, so stop was moved to breakeven.   The market rallied   back  up,  just touching my stop, taking me out at breakeven before a   further   sell-off to 2213.25 then another rally.

On the plus side it was a very good entry point, 2 minutes after the high of the day session.

Down 0.25 points (after commissions).

I  came back from a break to eat to see I'd missed a sell signal  at   19:52, although I had my laptop with me I now realise that if I  unplug  the  ethernet cable, although it does switch to wireless, when I  go  back to  ethernet my indicator stops updating but the NASDAQ  doesn't....no  idea why that should happen but definitely something to   remember.

No drama since although the market sold off from  2217.50 at 19:52 down   to as low as 2209, it then rallied back up to  2221, if  traded, it'd  given the same result as the first  trade, flat (since I  move stop to  breakeven if the market goes 5 points  in the direction of my trade).

End of the black line on the top chart marks the second sell signal point.

Remember folks;

"Analysis without trading is like foreplay without sex"

Monday, 21 November 2011

I can only trade the US afternoon session this week. 

There was a beautiful buy signal earlier on, coming just after the low of the day (so far at least);

However there was also a sell signal, in my absence, at a less than favourable time;

Friday, 18 November 2011

[quote name='PositiveDeviant' timestamp='1321639427' post='232326']
There was another potential signal at 17:11 however it seemed marginal whether it was a valid signal or not, and it was quite similar to signals I took on my old strategy, and I'm deliberately avoiding those as they tended to result in 50/50 probability trades;

This  is important since the trades using my previous strategy (double   tops/bottoms on my indicator) were not as high a probability as those   using my new stratgegy, so I need to properly differentiate between the   two, when trading.

I went back through my trade data from when I  was trading between 1st  March and 8th August, but re-assessed it to  check what impact would be  had by;

Using a 10 point profit target and stop moved to breakeven after 5 points.

It  turns out 52% of the signals would have resulted in a profit, not  much  different to the 50/50 using the 8 point trailing stop strategy.     Previously, only 61% of the signals reached the 4 point mark, whereas   with the new strategy 72% of the signals reached the 4 point mark   (during the backtest).

Previously with ThinkorSwim the data was not good, so in order not to  be confused by many false signals I had to  set my indicator to a 5  minute timeframe, effectively meaning I was  always several minutes late  getting into the trades.

In order to  properly define what constitutes a signal I'll need to go  through the  full data set for the back test with my new strategy.

So far it  seems reasonable to state that the diverging higher  high/lower low in  the NASDAQ needs to be at least 1.5 points beyond the  prior  high/low, and the PositiveDeviance Indicator needs to display a  divergence  the opposite way by 0.040.   Also for the backtest I was  using a 3  minute delay following the production of the high/low before  the trade  was classed as having been taken.  I may shorten that to 2  minutes since  the data is of a much higher quality now.  I may also  annotate the  charts more during trading, as part of the process leading  up to getting  into the market.

PDI point to point >0.040

[up to MT DBDT DIVERGENCE 16/09/11]
Buy signal at 14:55, filled at 2262.50, market rallied up to 2268, moved stop to breakeven, market then came back taking out my stop at breakeven.  The point where I bought was just prior to a difficult choppy phase...;



There was another potential signal at 17:11 however it seemed marginal whether it was a valid signal or not, and it was quite similar to signals I took on my old strategy, and I'm deliberately avoiding those as they tended to result in 50/50 probability trades;

Thursday, 17 November 2011

Had a buy signal shaping up for 16:26 however my indicator formed a further low at the same level, deferring the signal to 16:30, by this time the market had rallied 9 points from the low therefore I opted to pass on it, the probability for a further 10 point rally on a downtrend day appeared remote.  As the market rallied just over 5 points above the signal point before selling off, it'd have resulted in a breakeven trade.

Buy signal at 17:45, filled at 2264.75, market ralled up to 2269.50, moved stop to 2264.50.  This signal came just following a 15 minute, 28 point drop in the NASDAQ, I was looking for a very sharp short covering rally, but it didn't happen, market sold off down to 2258.75 on the next move down, taking me out 2264.5 for a 0.25 point loss. 

Buy signal spanned several days;

A nice clean buy signal this time.

Entry and exit;

In hindsight although the buy was 4 minutes after the low of the day, the market didn't stage any meaningful rally.

Down 0.5 points (after commissions)

Wednesday, 16 November 2011

The NASDAQ just cratered there;

Something to do with this perhaps?

"Fitch: U.S. bank outlook could worsen over Europe

Stories You Might Like

By Wallace Witkowski SAN FRANCISCO (MarketWatch) -- Fitch Ratings said Wednesday that the credit outlook for U.S. banks can worsen if the euro-zone debt crisis is not resolved in a timely manner. "Fitch's current outlook for the industry is stable, reflecting improved fundamentals at most banks combined with ratings lower than at pre-crisis levels. However, risks of a negative shock are rising and could alter this outlook," the ratings agengy said. Fitch maintained Europe's sovereign debt crisis still poses a threat to U.S. banks even though the institions have reduced their exposure to the region over the past year."
Had a sell signal at 19:45, filled at 2356.75, the market sold off down to 2351, moved stop to breakeven. The market then rallied back to 2355.75, then started selling off again, down to as low as 2334 as I write this, therefore my limit order was taken out at 2346.75 for a 10 point gain.

A gain of 8 points today overall (after commissions)

Mutlicharts issues...

Had some more problems with Multicharts today, I had a buy signal at 15:37, filled at 2349 however Multicharts didn't send my stop order into the market with the opening order, nor did it show the orders on the chart, it was set to do this but for some reason it didn't appear.  Due to this I closed the position immediately, for a 0.25 point loss.  Back onto technical help, they advised to reset broker settings (symbol mapping), we did that, then I sent an order in to short with a 1 point stop to test it, it worked this time, and also the issue with the stops/bracket orders doubling the number of contracts seems to have now been resolved.  The test trade got stopped out for a 1 point loss.

The original trade would have been stopped out for a 3.25 point loss, had I not had the technical issue, so it actually saved me 1.5 points.

Down 1.75 points (after commissions).

First trade entry (marked manually), and second trade.

Tuesday, 15 November 2011

Currency divergence with E-Mini NASDAQ, EUR/USD, AUD/USD and AUD/JPY (19th Aug to 15th November);

This shows clear divergence between the NASDAQ and key currencies between 7th and 21st September,  NASDAQ trending up whilst currencies are trending down, this tends to lead into corrections in the NASDAQ.  More recently the Euro is diverging away from the pack.

Currency divergence with E-Mini NASDAQ, EUR/USD, AUD/USD and AUD/JPY (21st Oct to 16th November);

Looking more closely, recently there is clear divergence of the type that leads into a correction, similar in nature to the first chart between 15th and 21st of September, but not as pronounced.  I understand EUR/USD is now at a 5 week low.

There were lots of decent moves today but I didn't get any signals to trade any of them.  Went short at 15:50, filled at 2342.50, market sold off down to as low as 2335.25, market then rallied taking me out at 2342.25, leaving me flat for the day.  Same issue as yesterday, Multicharts sending twice as many contracts for stop/bracket orders. I've upgraded to the latest version of the software so we'll see what happens next time I trade.


Monday, 14 November 2011

Four trades today;

1st and 2nd trades

I prefer to have a grey background these days so hopefully it's ok on your eyes folks.  The arrows at the top/bottom of bars show the bar where the entry/exit took place, and the arrows to the side show the price level where the trade occured.

Went short at 14:45, filled at 2347.75, market sold off down to 2339.25.  Since one of my new rules is to move to breakeven if the market moves 5 points in my favour, I moved my stop to 2347.75.   Something slightly unwieldy happened when I placed the initial order.  I have Multicharts set to automatically attach a bracket OCO order.  So a fixed stop of 3.25 points along with a limit order to take profit at 10 points.  For some reason Multicharts sent the bracket order into the market with double the number of contracts, and when I tried to amend the order it was rejected.   It just meant when the position was stopped out, I also went long by one contract that I didn't want, and just closed it immediately, for a 0.25 point gain.

Next signal was short at 15:17, filled at 2357.25, the market then sold off down to as far as 2344.25, hitting my limit order in the process, for a 10 point gain.   When I initiated this 2nd trade the same issue with the bracket orders occured, I just had to immediately cancel the bracket orders and re-enter the stop and limit orders manually.

I got through to Multicharts technical help, the guy didn't provide a clear answer as to what may have happened.  I can only think I had two separate bracket order strategies on auto-attach, but he seemed to think that would result in the stop price and limit prices doubling, not the contract size.  He was quite helpful though, and put a test order into the market many points away fom the price, and the bracket order attached properly after the settings were reset.   There was also a buy signal at 15:52 (while I was on to technical help), that would have been stopped out at break-even anyway.

Third signal came at 19:12, went long at 2339.25, market sold off taking out my stop in the process for a 3.25 point loss.

Fourth signal came at 20:04, went long at 2334.75, the market edged higher, towards the close it appeared increasingly unlikely that the target of 2344.75 would be hit so I manually trailed the stop higher, before it got taken out at 2338, for a 3.25 point gain.

3rd and 4th trades

A gain of 9 points today (after commissions).

The 2nd trade was also signalled by my indicator on a larger timeframe;

Sunday, 13 November 2011

This is the MFE of each signal in ascending order ;

There are only 14 signals that achieve a better outcome with an 8 point trailing stop effectively showing that the 8 point trailing stop strategy is a more risky proposition than a fixed 10 point profit target strategy.

Saturday, 12 November 2011

After doing some more tests today I'm going to have to go with the 10 point profit target strategy.  Since I can't trade full time (yet) I looked at what happens to each of the two strategies when they are traded on and off, as in, one week on, one week off (as I'll be doing).  The 8 point trailing stop strategy suffers quite a bit, since a relatively small number of the trades account for a large share of the gains, whereas using a 10 point profit target strategy means the profits are spread more evenly out accross all the trades, meaning less impact from on-off trading.

Tuesday, 8 November 2011

What I want

I've now been through signals from 8th September 2011 to 31st October 2011.

71 signals

39 wins (55%)

20 losses (28%)

12 breakeven (17%)

72% of the signals reached the 4 point mark also, similar to before.

The percentage splits continue to be similar, so far this is good as it suggests consistency which is what I want, especially the 2 wins to each loss.  That suggests a decent edge but I need to go through more data, and on more markets to check it properly.  It's quite time consuming to make sure it's done objectively.

I ran a plot of the cumulative points using this new method.

Sunday, 6 November 2011

On 2nd November I described a new way that I can use my indicator that looked quite interesting.  I've been through data on the NASDAQ from 8th September onwards and highlighted all trade signals on it using this new method.  It's here, if anyone's interested.

Essentially, when my indicator deviates from the NASDAQ, I trade in the direction of the indicator, as it can point to short term changes in trend.  In light of this I've re-named my indicator the Positive Deviance indicator.

I've been through the signals from 8th September 2011 to 6th October 2011 and the early results look interesting.

37 signals

20 wins (54%)

10 losses (27%)

7 breakeven (19%)

If traded, the NASDAQ points gained on these signals would be 162.5 (net of commissions), for this one month period.  This equates to 135% return on capital risked (37 X 3.25 points risked per trade).

This is working on the basis of using;

A 3.25 point stop, as before
An 8 point trailing stop, as before
And moving the fixed stop of 3.25 to breakeven if the trade moves 5 points in my favour.

During my period trading from 1st March 2011 until 8th August 2011, I kept statistics on trades showing whether the trade moved at least 4 points in my favour, from the point when the trade was initiated.  Based on the signals that were traded, 56% of the trades reached the 4 point mark.  Also the signals were 50/50, 50% would result in a loss and 50% would result in a gain.  

So far, using this new method, 70% reached the 4 point mark, and so far the winners are 2 to 1.

Although it looks interesting, It's not really wise to draw any firm conclusions on the basis of such a small number of signals, once I've gone through the rest of the data tomorrow I'm looking for it to be consistent with the above.   I could then also look at other markets (other than the NASDAQ) to see if similar results are seen, then I'll have a better idea if I've created a true edge.

Wednesday, 2 November 2011

Is this the second part?

I'm now set up on an internet connection that is more than twice the speed it was before with a much faster ping - much better for trading obviously.  I now get 124 ms to Chicago, an improvement on the 168 ms I had before.  That small amount should result in improved fills - the faster the better.

I'm pulling together a number of different elements before I restart trading.  That's another ticked off the list.

This break in trading was, at least initially,  forced on me by ThinkorSwim ceasing operations in the UK.  It's really turned into a blessing, allowing me to re-asses different aspects: broker, internet connection, trading plan etc

I've been spending some time thinking about some of the guidance Jeff Quinto gave me and I'm still working on what the conclusion to that it is, and how I then integrate that into my trading plan.  It makes sense to cut costs in my trading business, and one way I can do that is by reducing stop size.  One of his statements;

"A trade should work immediately and profoundly"

For each winning trade I had previously I looked at the number of points the trade went  against my postion before going in a favourable direction and turning  into a winning trade. From those statistics, I found the average stop requirement for winning trades was rather low - 1.29 points. The average is of course just that, some went 3 points against me before working, others just immediately worked and therefore bear out what Jeff is saying, the best trades work straight away.  I's quite tempting to simply reduce my stop size to 2 points (from 3.25), this would mean, more stopped out trades, but it would also mean no hanging around waiting for a trade to work.  If a trade doesn't work within 2 points of forebearance then I will be quickly taken out of the trades that aren't winners, and left to focus on how I manage those that work immediately and profoundly, as they still have room to flourish. 

This also follows the logic of the "Phantom of the Pits", writings by a former floor trader.  His rule 1 is that you wait for the market to prove you are correct rather then wait for the market to prove you wrong.  Many traders put a position on with a stop.  Their view of the position is that if the stop is hit, their position has proved to be wrong, and therefore they have exited the market.  There is a subtle but interesting difference between waiting for the market to prove you correct and waiting for it to prove you wrong.  Waiting for a position to be proved wrong is a focus on the negative whereas waiting for a position to be proved correct is a focus on the positive.  You can only be waiting for the market to prove a position to be correct if you decide that you are only interested in positions that work imediately.  And if you decide you are only interested in positions that work immediately then you absolutely must either - reduce your stop size, or - exit postions very quickly if they don't work straight away.

Of course that is only one approach, another method would be to take a more different approach to trade entry.  Previously my rules for trade entry were very simple.

1. Signal on my indicator - check
2. Market having moved in the direction of the signal for at least one point during the last one minute bar - check
3. Trade order sent.

That was it.  The positive from that was that there is never any question or doubt about getting into a trade, I only needed to check 2 points.  The simpler it is, the easier it is to execute efficiently.  However, by waiting for the market to move 1 point in my favour, that means I'm losing ground of at least 1 point on a lot of the trades before I even get in.

Another approach may be so look at the average "noise" produced by the market on any given day.  The market always has an element of 3 steps forward, 2 steps back, and what I will look at doing is taking advantage of that, perhaps by looking at Keltner channels on a very small timeframe.  That way, I could enter postions when the price is at least in the opposing half of the Keltner channel, before initiating a trade, that way I may be able to pick up on that 1 point, previously lost on many trades but still capitalise on the "trades should work immediately and profoundly" aspect.  It may allow me to enter the same positions, with 38% less equity risked for a similar outcome trade.  It may also lead to some missed trades also though, so I need to study that further.

Ive also been looking at the approach I'm taking with my DBDT indicator, I've tried layering it on top of a chart of the E-Mini NASDAQ (NQ) but it looks quite difficult to read.  I've had a persistent interest in why double tops or bottoms tend be a good signals on my indicator.  What I mean here is that I tend to get a double top on my indicator where there is no double top on NQ, so what is actually happening there?  I've speculated previously on what might be causing that to happen yet hadn't felt close to an answer.  Today, I feel much closer.  It's divergence.  If my indicator produces a double top when the NQ is showing a higher high, then my indicator is diverging in a direction opposing the trend in NQ, hence being classified as a sell signal.

The picture becomes far clearer when I place my DBDT  indicator chart side by side with the NQ chart.

So, the top half of the chart is the E-Mini NASDAQ futures (NQ), the bottom half of the chart is my DBDT indicator.  For anyone new to my blog here, just look at my indicator as my personal re-valuation of the NASDAQ.

Now this is an interesting chart.  What I've shown here is that by looking at my indicator where lows or highs meet, there are occasions where these diverge from NQ in the direction the NQ is about to travel.

Point 1, it's clearly seen that this is a higher low on my DBDT indicator but a lower low on NQ.  Shortly after the production of the low, the market then travels in the direction of the DBDT divergence - up.

Point 2, my DBDT indicator is flat whereas as NQ is higher, DBDT diverges away from the direction of the trend, and NQ follows thereafter.

Point 3, a lower low in NQ with a higher low in DBDT, again the market travels in the direction of divergence on DBDT.

Now let's head back to Ray Dalio - Is it true?  

Well, it's certainly true of today.  Divergence at Point 2 was the high of the session today for NQ.  Divergence at point 3 was the low of the session today.  So today, divergence on my indicator has pointed to a probable direction of the market in NQ.

Now you can well imagine that if this type of event is a regular occurence then today I have found an altogether new and very interesting direction to take my trading in.  Not limiting my trading to double tops or bottoms on my indicator but expanding it to trading these types of divergence events will make sense if I can prove that this isn't simply a one-off.  In fact I'm going to make a concerted effort to try and prove that this is wrong.

The market yesterday, 1st November;

Similar events....

31st October;

Again (!), similar events.

It's getting late, I'll need to look at this very closely tomorrow.

28th October 2011;

And again....

I remember some time ago, I think it was not long after I came up with my indicator, I had a dream I was talking to a world famous investor and he told me, "You need to look for the second part".  I was completely flummoxed by that at the time. 

Have I just found "the second part"?