Continuing issues with NT this week--in this case connection problems (see thread elsewhere) led to creation of a phantom position in NT's record of my live trading account. In this case a position that shows up nowhere else except as a position in the Executions tab and which is used internally somehow in P&L trade calculations every time a trade is closed, totally screwing NT's accounting prior and since. As usual NT refuses to sync with the broker account despite restarts, DB resets etc. ad nauseum. Reinstalling NT (and losing all records kept) is only cure I've found so far for this particular problem.

The good news is I'm learning my way around the TWS trading interface -- ugh.

ETA: in other news took a hint from Big Mike's inquiry in another thread about value areas and started looking closely at the "volume profile" stock indicator as a constraint for trading what the system otherwise interprets as chaos (i.e., chop), especially for predicting jumps between "modes" (or areas of concentrated activity). These areas so far seem to correlate with a crude realtime activity indicator (essentially, activity = 1 / [seconds per tick bar]). Not exactly sure what volume it is measuring in the case of Forex (appears to be at least what is relayed from IB as instantaneous inside order sizes) but if the last couple of days are indicative the approach may prove profitable.

Trading activity described today in this post. Bottom line is I scalped 2 contracts all morning (80 trades, all in at entry with max 1.5 pip stop + 2 targets) and grossed ~73 pips, netted ~33 pips after commission (bandit of a broker made off with ~40 pips).

Traded from STF (200 tick) chart using price action and stochs, MACD and VolumeProfile in order of priority.

NT accounting is still FUBAR'd and will likely reinstall it over the weekend.

ETA: I've copied the chart image I posted to the thread alluded to above in order to comment briefly on it. I think a lot of traders would agree the style of trading indicated, while profitable at the end of the day, is very inefficient in terms of commission costs. However, it reflects something I've found to be true; namely, if one intends to trade (and I wake up intending to trade every morning the market is open) one probably ought to trade one's mindset rather than one's system. This means in particular that if the first 3 trades go against me that's it for the day. Otherwise if up to it and the market is cooperating (i.e., relatively large moves in response to events, say, clear sentiment, obvious signals) I'll start trading the "5 energies" system described previously. However, some days (like today) all I can focus on is minutiae of price movement, and those days I scalp, more or less trading the activity histogram of the current mode, depending on MACD for temporary sentiment, stochs for entry, entries and targets at local S/R levels , hopefully catching the trend when price migrates to (or establishes) a different mode but often as not scalping every wiggle along the way.



Detail of 200 tick chart showing what I mean by "scalping every wiggle" . With NT accounting misbehaving I'll have to wait for the broker to post the activity report to calculate trading stats (other than the loss to commissions) for the morning.

Mixed results Monday & Tuesday, partly because I broke a rule that because of it's apparent importance probably ought to move up the priority list; namely, don't trade outside your normal session. I've been trading outside the normal session (combined US/London session) for 2 reasons:

1. I was trying to look busy to avoid doing repairs to the deck at the back of the house; and,
2. I was curious whether at this stage of my trading career the scalping method used recently to trade the regular session (both New York and London online), when trader activity is 10 to 20 x activity of New York alone after the London close, would work in a slower market. The question occurs to me now and then because range and tick bar patterns have no time characteristic and the 2-part hypothesis to be proved or disproved is therefore that

1. patterns are all that matters.
2. patterns don't change when activity slows

This time around it turns out the approach (setting out to make a number of relatively small, short time frame trades, essentially scalping) did not work outside of normal session hours, partly because the slow pace drives me absolutely nuts (nuttery in this case characterized by a compulsion to gamble). Thus we chalk up another statistic on the "How not to trade" side of the ledger.

The last few days reminded me of trap I sometimes still fall into stems from the days I tried to predict what price was going to do in the hope of getting in at the earliest possible moment. Stats continue to suggest the odds of doing this (in my case) are less than 50%, probably much less. Worse, stops and targets tend to reflect expectation of the Big Move and hence are in general wholly unrealistic in light of what price is actually doing--all in all a recipe for disaster.

In an effort to regain the almost 50 pips lost so far this week today we're setting out to trade the mode a little more objectively. The figure below shows a 4-standard deviation regression channel (from a couple of hours before the N Am session started and the present) added for guidance and order preset at the top of the channel (top because of the slightly negative slope of the channel). Ordinarily I don't preset orders but wait for signs that a retrace is setting up, and intend to when the time comes. Being easily amused however, in the meantime the preset gives me something to occupy myself with (a point of reference for ongoing price action), which prevents me randomly clicking Buy/Sell buttons



ETA: In as much time as it took to write the first installment the setup was triggered with the following outcome:

First target hit, ATM strategy advanced the stop to breakeven + a couple of pips.



Second target hit.



Subsequent price action. Top of regression channel (blue line at very top of chart) now indicating mode bias is upward so looking for a long trade once robust (upward) movement sets up below the mean/median of the mode



Next 2 trades (first long at regression channel bottom wary of aggressive downward momentum, target 1 + trailing stop hit; second short in case transition to new mode occurring--target 1 + trailing stop hit.

Hate trading the thin edge of the wedge, especially around MM 0/8 ("Ultimate S/R"). Wouldn't hazard if any powerful magic numbers (00, 50, 75 or 90, say) in the vicinity




Final couple of trades of the day, both short: first non optimal for 1 target hit + trailing stop; second actually after London closed but higher probability, by preset ATM strategy while I wasn't in the room--both targets hit

Generalities

[Started much earlier today] Trading the European open just now (not my cup of tea ordinarily since the price action lately rarely seems to justify being awake in the wee hours AST) reminds me of the discussion about chop, how to detect it & how to deal with it. I define chop as price action with respect to some time frame that doesn't generate profitable trading opportunities within the confines of one's system, the only gauge that works for me being that if I try to trade it and get whipsawed then it's chop (no longer whipsawed to death, fortunately, since these days I back off and reassess the first time a trade doesn't go as expected).

What has worked for me more or less so far this morning (spot EURUSD, 8 trades, 6 profitable, 2 breakeven) in the absence of anything I'd call a trend continues to be treating the position of price relative to VolumeProfile clarified by a 4-stddev regression channel as an indication of the probability price will move in a given direction (both indicators applied to the European session only, assuming there is going to be a disconnect between European and Pacific Rim trading patterns), further constrained by various S/R levels (Murrey Math, PriorDayOHLC) and by momentum in 3 time frames (200, 600 and 1800 tick chart).

The "edge" here, if we can call it that, derives from the awesome power of not trading (similar to the power of quitting, just that you can "not trade" as often as you like whereas presumably one quits only once per session ). Price can fluctuate at random between 4-stdev regression lines for a relatively long time before approaching one boundary or another, at which point, and only at which point, any potential trade is considered. Others use some mix of other volatility based channels (e.g. Bollinger Bands) in a similar way (to estimate the probability of a reversal versus a continuation or "breakout"), and some even have strong opinions and elaborate theories explaining why it works. IMO the bottom line is one ought to go with whatever works without necessarily trying to explain to anyone else why it works (and especially don't try to argue one technique is somehow superior to any other since such explanations simply cause the listeners' eyes to roll back in the head. IMO all techniques that consistently make money for the practitioner probably work because they've become part of a given trader's wiring, no more and no less, exactly the same in light of the only metric that matters--profits). That said, I suspect keeping an eye on VolumeProfile and regression channels on top of everything else works for me simply because the additional constraint biases against over-trading--invariably when I'm not at the top of my game--which still remains the occasional impediment to consistent profitability.

Upshot--the power of not trading in action

[At end of day] Having said all that I finished today one pip in the black after 42 trades, ca. 60% profitable (which stats may suggest what kind of day it's been)--a miracle, having missed all significant moves getting what little sleep I've had in the last 24 hours while they were occurring and eventually making all the mistakes I've been pontificating against . What may have saved my bacon (like landing a Cessna 170, any session I can walk away from is a success) was somehow managing to recognize when price pattern was shifting, working with tiny pip targets & microscopic stop loss. That and the triple fail safe--I've finally accepted that if I begin to feel anything at all, especially negative emotions, it's a huge red flag. Not necessary or even possible to control them at that point but essential to stop trading.

Tomorrow is another day and TGIF

bnichols,

Just wanted to say I've been following your journal and enjoy it a great deal.

bnichols 

What has worked for me more or less so far this morning (spot EURUSD, 8 trades, 6 profitable, 2 breakeven) in the absence of anything I'd call a trend continues to be treating the position of price relative to VolumeProfile clarified by a 4-stddev regression channel as an indication of the probability price will move in a given direction

Hold on, if you're trading spot EURUSD, how are you getting volume profile? Are you taking it from the EUR future?

Adamus 

Hold on, if you're trading spot EURUSD, how are you getting volume profile? Are you taking it from the EUR future?

Hi Adamus !

Roger that--good point and alluded to briefly in post 71 above, VolumeProfile plots whatever is returned by NT for spot Forex as MarketDataEventArgs.Volume in the OnMarketData event, which for Forex data may be what NT represents as instantaneous inside bid/ask volume in the lower right hand side of a chart (Chart Trader section) and for IB in particular (sampled data) is in any event highly suspicious (but which nevertheless, e.g., jtRealStats manages to create a volume ladder out of). Here we're interpreting whatever it actually is simply as a measure of market activity at a given price.

By the way, I know I'm giddy from lack of sleep when the question brings to mind the scene in HitchHikers Guide to the Galaxy where Jeltz (in charge of the Vogon demolition fleet) responds to the panic of the population of Earth when they learn the planet is seconds away from annihilation, with "There's no point in acting all surprised about it. The plans and demolition orders have been on display at your local planning office in Alpha Centauri for fifty of your Earth years, so you've had plenty of time to lodge formal complaints.''

B.

mokodo 

bnichols, Just wanted to say I've been following your journal and enjoy it a great deal.

Many thanks mokodo !

Friday was a learning experience, starting badly (down about 36 pips or $360 after commission at one point) but ending on a positive note (up about 42 pips or $420, only because the market took pity and delivered almost $750 in 16 trades over the last 42 minutes of the session), 67 trades in all @ 57% winning, 2 contracts per trade, 2 targets, 1 contract per target.

What today made clear was that if I choose to trade like a rodeo cowboy (so to speak, meaning ignoring the rules & trading by the seat of my pants when there are no clear signals) I'm going to faceplant often & eat a lot of cow pie.

A trade-by-trade post mortem of the week is still in progress, which includes a comparison of trading results to backtest results for a trend trading bot (see attached strategy cobbled together this A.M. based on the anaSuperTrendU11 indicator)

BUT NOTE IF YOU WANT TO TEST THE STRAT: the strat calls an experimental version of anaSuperTrendU11 (not included since it's [MENTION=3534]Fat Tails[/MENTION]' IP) that has been modified simply to give direct access to the UpTrend[] dataseries; specifically the scope of the UpTrend[] dataseries has been changed from private to public on liine 75 of the original; i.e., from

 

Code

private BoolSeries upTrend;

to

 

Code

public BoolSeries upTrend;

Once optimized on 1800 tick data for the last week no surprise the strategy traded a heck of a lot less during backtesting (averaging 1 trade per day, 80% winning, hence suffered much less loss to commissions) and doing significantly better in terms of P&L than I managed when a usable trend emerged on Friday.

Results for Friday


My trades vs bot trade on 1800 tick chart:


While the bot made $721 on 1 trade (2 contracts) and paid $9.92 in commissions, my trade-by-trade P&L looks like this:


Notes on attachment

Most of us know what bots are worth (anybody watching the machinations and share price of Knight Capital Group?). Guessing which is why high frequency bot programmers earn the big bucks, but live and die by the sword.

This is a barebones trend trader entirely dependent on Fat Tails' anaSuperTrendU11 indicator. To use as is you must modify a copy of the indicator as mentioned above.

To use with unmodified anaSuperTrendU11 change the existing OnBarUpdate() code below in TDSuperTrend.cs to detect a change in direction from an available anaSuperTrendU11 dataseries (i.e., StopDot or StopLine)

 

Code

protected override void OnBarUpdate() { bool isUp = anaSuperTrendU11(BasePeriod,Multiplier,RangePeriod,ReverseIntraBar,ThisBaseType,ThisOffsetType,ThisRangeType).upTrend[0]; if(isInit) { bool isGoLong = isUp && !isLastTrendUp; bool isGoShort = !isUp && isLastTrendUp; if(isGoLong) { string sigName = "L_"+nTrade.ToString()+"_"+nLong.ToString(); nTrade++; nLong++; isLastTrendUp = true; EnterLong(Quantity,sigName); } else if(isGoShort){ string sigName = "S_"+nTrade.ToString()+"_"+nShort.ToString(); nTrade++; nShort++; isLastTrendUp = false; EnterShort(Quantity,sigName); } } else { isInit = true; if(isUp) isLastTrendUp = true; else isLastTrendUp = false; } }

It seemed so much simpler this morning to modify the indicator and thus inherit Fat Tails' impeccable logic than to reinvent the wheel in the strategy.

In any event if there's an issue let me know. The only claim to fame for anyone who doesn't already have a method in their own code snippet collection may be that the strategy shows one way to expose an indicator's enums to optimization.

bnichols 

BUT NOTE IF YOU WANT TO TEST THE STRAT: the strat calls an experimental version of anaSuperTrendU11 (not included since it's [MENTION=3534]Fat Tails[/MENTION]' IP) that has been modified simply to give direct access to the UpTrend[] dataseries; specifically the scope of the UpTrend[] dataseries has been changed from private to public on liine 75 of the original

Although upTrend is in fact private, the anaSuperTrendU11 has a public BoolSeries UpTrend defined in line 564 of the Properties section. You could have easily accessed the public BoolSeries as anaSuperTrendU11.UpTrend[0] without modifying the indicator. Please take note of the capital "U" of the public BoolSeries.