Clemmo's Trading Odyssey

Tear It Down 2

Dissatisfied with the results of the simpler idea, I added complexity in an attempt to 'improve' it. I did not succeed.

“Just when I thought I was out they pull me back in.”
Sonny Corleone

Principles
Simply, as soon as a big juicy bar goes up we expect price to retrace at some point to at least ~50% of that bar.

Indicators
BarTimer to measure time left in current bar.
SpikeBars 3.2 (paints the bar after close)
ThrustBars (signals immediately at threshold and paints the candle but will remove the flag if closing price is below the threshold)
Combine Thrust Bar & Spikebar indicators to flag and signal all big bars.
Jotun SMMA
Use TMA envelopes & TMA slope to show where non-prodigals are most likely
Use TMA slope to help you decide when to let some trades run? Not this system.

Add and tune templates for as many instruments/TF as CPU can accommodate. Scan pairs, on M15/H1/H4(?)/D1(?); trade ones that aren’t correlated.

Entry:
Enter immediately after 80% barTimer on any signal that is opposite to what Jotun is flashing UNLESS it’s near the band edge.
Enter immediately after 80% barTimer when Jotun is candy-caning (ranging).
Do not enter when:

1. Price is touching the band edge/outside the bands and signal is towards the reversal direction. So if price is above the band, and you get a bearish tall bar, go for a trend-trade instead or ignore.
2. The candle is 200-250% taller than the mean average and doesn’t retrace immediately. These are trend signals. (mega-bars)

Exit:

1. If the retrace hasn’t happened within X time just cut it loose and go for the next one, there will be no shortage.
2. Two tall bars in a row is a trend - exit and reverse and hold til band edge.
3. If an adverse excursion hits the wrong side of the band
4. When multiple spike bars form (invalidating your retracement trade) you have to ACT FAST to get in heavy with the trend to recover from the drawdown. (often fatal)
   

Maximum Wait Expectations for retrace - still anecdotal and instrument-dependent but so far I would say (EURUSD example)
M15 - 10 pip bar - 20 hours
H1 - 20 pip bar - 1 day (24 hours)
H4 - 25 pip bar - 4 days
D1 -125 pip bar - 20 days
If price hasn’t retraced within that time abandon the trade as hopeless.125 may be too tall. 100 looks just fine and provides more trades.

Stop-loss idea - no more than 1 (or 2) non-prodigals of the same direction at any time.

How do people trade manually? Seriously asking.

It's just so dull.

New York opened and the pound is just. sitting there.
No big news on the calendar. No one's waiting for something to happen. It's just doldrums.

It's dull, except for the nerve-wracking situations that happen all too often to most: A trade is running that started off going according to plan (really, for most according to a guess), makes a retracement that keeps going deeper than expected, leading to "Should I just close it now with a small loss, or stick to my plan?". And then it turns around and resumes course--"Great! Sticking to my plan was the best decision!". But then price goes onto another retracement, deeper and deeper, nerves become jangled, price falls more, but "I stuck to the plan and it resumed earlier, so I'll continue to stick to the plan!", and now so far in the red that don't want to take such a loss and just hold on, stomach churning, brow dripping, as the trade turns redder and redder.

Often if you look hard enough, you find something to do.

Yes, if you look hard enough.

I guess I should have expected some consolidation after the barnburner on Friday.

Top Top Top Top Top Top TOP Top Top Movers

https://docs.google.com/spreadsheets...YWY/edit#gid=0


The 'top movers' experiment continues and this is a reminder to myself to revisit it every month at a minimum; although based on what I see the conclusions are already fairly clear and support what I already believe (confirmation bias!), namely that momentum (in the physical sense of something that keeps going once it's started and is hard to get going if it isn't) exists. The one thing that is curious is that if something is really inert, it's quite likely to get a jumpstart. The market makers cannot tolerate a dead instrument. They need prices to move to take their cuts.

The project name comes from a recurring gag in a hilarious podcast I was introduced to. If you like dry British humour you should love this.

Krunk-O-Matic

Scalping notion

Basic idea: Trade any (two?) three pairs that will somewhat self-hedge, for example, as Tokyotrader suggested, GBPCHF, EURUSD, and USDCHF, but there might be better options - do some research.

Use Tipu Trend panel to tell you when these currencies are all trending. Not ranging.

Buy or sell all pairs, maybe in unequal lots to balance trendiness?

Use the trump-closeout EA to close all trades once a profit target has been reached. If profit target hasn’t been reached by the time one or more pairs starts to range again, close out. Maybe close out only the ranging currency, the unprofitable, the profitable - lots of testing required.

Repeat.

Anecdotally, this has been fairly successful.
But then why didn't I continue the experiment and record the results more thoroughly? It must have gone off the rails at some point. I now know after reading Parry Kaufman's book that this is basically a pairs trading idea and that we can theoretically improve the hedging by adjusting the trading volume to correspond with the volatility of each instrument. I'm not really sure if this is a system so much as it is a method of selecting and sizing trades.

FX Meter Contrarian

I no longer recall which FXmeter I was using - maybe Abokwaik's?

A strongly moving pair of currencies on a lower TF (M5/M15) that is NOT also trending on higher TF may be a sign of a generator phase being tapped out and price ready to resume with the main trend (the initiator). Other signs include a large spike bar/ATR flag.

This system simply places trades by doing the opposite of the recommendations on the M5/M15 timeframes of a currency meter app. Possibly bolstered by a quick glance at a chart to determine if it really is moving against the main trend on a higher TF.

Ideas: similarly, signals on the higher TF (MN, W1, D1, H4) may be due to retrace and might offer some contrarian opps.

Initial testing positive/mixed - 4 different weak/strong pairings and 3 different scenarios. One ranging, two strongly trending, one reversing.

There are a few problems with this notion. First of all price mean-reverts a lot more often on lower TF; seeking higher TF reversion trades is not time/cost effective. Secondly, if price hasn't been trending on higher timeframes, it's likely to keep on ranging even if it gives you whippy trendy signals from time to time on lower TF. It depends on how long it's been stuck in accumulation/distribution. If you're looking for breakouts, just use a straddle; and if you want to trade mean-reversion, you don't need a currency meter. Still, the idea that a strongly trending M1/M5 instrument within a range is likely to reverse (and even more so after an aggressive bar) is solid enough.

Contrary Canuck

Chart: USDCAD
Timeframe: H1
Indicator: ATR channel breakout x2

Is there any currency that is more irascible and grudging than USDCAD? (yes, probably; I'm not comfortable with the idea of an instrument having a 'character' unique to itself, except over short terms. It's all based on the same people/machines and what's old is new continually; having said that USDCAD and AUDNZD are ruffled similarly.) Every single movement is hard-won with multiple retracing, and doubling back. Most of the real movement in USDCAD is done in ‘slow leaks’ with gentle ranging movements that guide slowly one way then the next, politely respecting the ATR channel borders. Only rarely does this boundary get transgressed and when it does it’s an opportunity.

Is it a bit like the Canadian stereotype of the uncertain, demure Canuck who passive-aggressively feels the need to claw back every accomplishment of his powerful neighbour to the south? Or the hockey player who is polite enough off the ice but will take your head off at the first whistle?

Whatever it is, in something like 41/53 (recent) ATR channel breaks on H1, they all retraced to their origin point or close to it (75%). So the simple strategy is to wait for any ATR channel break (using our good ol’ double ATR-channel-break indicator) and then watch it carefully before, and just after candle close. Trade the other direction (fade it) once it starts to retrace and it will close the gap in nearly 80% of instances, and usually within a few (5-6) hours or the same day at most.

The other 20% of instances will be a big problem if you don’t monitor the situation. These are legit breakouts and if you get stuck in one of these, which you’ll recognize because you’ll get another ATR signal in the same direction, then exit, and reverse your trade x2 position-size for a longer haul. At which point the algos, sensing weakness, and desiring the recovery of 'their' monies will reverse on you hard.

You can sometimes avoid fading a legit breakout by observing the position of the ATR signal with respect to its channel. A breakout going short will start right near the top of the channel. A long one will start near the bottom. If the channel is downward sloping, any down break is more likely to be a real break, and vice versa. (Is it though?) So in those cases it’s probably better just to trade them like real breakouts, exit when price has crossed 70-80% of the channel width.

Despite my reservations, this will work. Even if you just stop trading after a suspected breakout. And don't come back in until the trend is tapped out. My natural impatience means this needs to be automated.

PriceForce Indicator (notes for)

Using analogs from physics.

Where mass = volume
Velocity = price change over time
Mass x velocity = momentum
Rate of change of momentum = Force

So,
ROC of Momentum = price Delta/time x volume
__________________
Time


We can achieve something like this using two moving averages as Hull explains in his book (must obtain) but can’t we find or make something more responsive that responds to fast moves and assumes they will produce the same degree of movement over double the time?
This treats market price like a physical system and it is not one. You can't graph the optimal strategy to a poker game. You have to react to highly specific circumstances as they come up.

Most satisfactory. 3 days to re-form the 'stabilizing honeycomb structure' (Neill) of a Wyckoffian range; and then the market simply took out the stops on both sides and powered up to confound the majority expectation of a retrace.

FX index plays matching it with some kind of 2D-3D OR 3D-5D fractals+making it skewed RR above cca 1:3 could work on all 8 currencies. Experimenting years.

this is the hardest part for me, actually ridiculus but really hard to achieve. And yes, you can have several systems with nice equity curve, nince mar ratio and everything...still founding out that you negative at the end of the road...its a curse of backtesting...as you get a quick result, while waiting for the market to move live

This is why I now only backtest to get an idea of market extremes and to confirm algorithmic correctness. And as for the extremes, I assume they can be 2x what I've seen in the history or worse.

Backtesting for optimization is fool's gold.

I haven't done one of these in a while; and I likely won't do another one.
https://docs.google.com/spreadsheets...xFY/edit#gid=0

This is a popular signals provider on many platforms. He is clearly good at business and has many social media channels. He brags about a 75% win rate for one of his EAs, (which means nothing anyway) but his forecasting signals aren't doing as well.

I checked his 30+ most recent trades (the smallest number you can check to get a significant result, so this is hardly 'proof') and of the forecasts that reached a completed state more than half are losing. What do you want to guess will be the ratio once they're all finished?

Of course getting a solid take on this performance is tricky because he often gives multiple levels, revises his older forecasts, uses a range of values or fails to provide a clear-cut target or stop-loss. All the usual tricks of the trade. One thing he does is to require price to move to a particular level before entry (good general practice) but then he often misses the entry and his target gets hit. I would say his directional bias is much better than 50%. This suggests that there is much more to being a successful trader than being 'right'. As you know if you follow the book club. What would happen if he stopped requiring price to meet him at a level? Well, he'd get fewer missed trades, but of the additional trades entered, there would be a corresponding ratio of trades that got stopped out immediately before moving on to their target. So the results would be much the same. In fact, if I count all of his 'missed entries' and 'first target' hits, (assuming that the second target would be set to break-even upon the first target being reached - advice he doesn't provide btw), then I get a ratio of 12 'wins' to 11 losses or 52%. If his targets are larger than his stops this is a profitable service. Is it worth paying following (these are free signals) for 52% accuracy? I won't try to influence your answer. Maybe his paid services are even better.

This is one of the top-rated signals/ideas guys on Tradingview and he regularly gets tons of upvotes and praise (possibly shills?) on his charts. This confirms both Clemmo's Conjecture and the wisdom of coppering the public's bets. If there is a successful signals service somewhere, they don't advertise, and have about 2 followers.

ps. Why do people always use chess pieces when they want to make something seem 'smarter' or more legitimate? Chess players don't know anything, except, obviously, chess.

yes agreed on the measurment of different scenarious.
while concerning optimization: i dont talk about fine tuning, i talk about obeying the system, not interfere; get used to 35-40% win rate, and waiting out the signals and trades
ridiculus-i find this harder than finding strategies with edge

how u guys tackle this?

Morrowind

Attached Image (click to enlarge)

Chart: EURUSD
Timeframe: H1
Indicators: 3 bollinger std devs; ATR channel breakout x2 (as per reclaimer)

There is an oddly discreet (did I mean discreet or discrete??) pattern in EURUSD - M/W/M/W that runs on a daily basis on EURUSD especially, but possibly some others (USDJPY?) and is easily visible on H1. It alternates with great reliability and I’ve only found a few instances where two Ms or two Ws run consecutively, although it’s not hard to make an M seem like W and vice versa. (and that's maybe one reason why this doesn't hold up)

The important thing is to note the ‘topological truth’ of the shape, not necessarily its actual resemblance to the letter. In that way, a W can start off with a bearish arm, as long as it retains most of the next ^ and hopefully the concluding portion.

This gives you a great deal of leading info but there’s still the usual problems with exits, trying to decide when a trend has ended, and spotting reversals.

Position -sizing: It’s not clear if some arms are consistently more profitable than others. Arms 2 &3 are the shortest and most volatile. You might want to put more weight on trades that match the higher TF trend.

There are 4 distinct moves per day, either down, up, down up, or up, down, up, down. This can vary considerably, especially during red news and on holidays with low volume. Consider avoiding.

You can possibly get some improved accuracy re: targets by using D1 as a higher timeframe guide. Basically, using reclaimer as the macro guideline.

It might be better simply to exit when there is profit and not worry too much about lost opportunity. Better to end trades early and be able to get in on the next arm at the start.

When you do make a mistake (miss an entry or an exit) don’t be in a hurry to close the trade at a loss, you might be able to make up the trade on another day, unless you are on the wrong side of a clear breakout.

The duration of each arm of the M or W varies and there is considerable room for error, some fuzzy logic will be needed, and pay attention to classic patterns like doji, engulfing bars, spiny backs, etc. (or don't bother)

I have very distinct memories of having marked up many months of chart data with the M and W shapes on EURUSD but now I cannot find the templates or the screenshots for any of it. WTF. Bad archivist! Bad!

But it's easy to make a new one. Look. I picked a day completely at random - it's clearly an M.

The next day?

Attached Image (click to enlarge)

Evidently, a W but the last arm of the W leaks into the following day. Is this a problem? Yeah, of course it's a problem.
We could skip the rest of this day and look at the next one. Should be an 'M'.

Attached Image (click to enlarge)

Is it an M? I guess? This gets very subjective. We need some guidelines. Luckily I made some.

Here are the general guidelines I have so far.

When drawing your M or W, try to connect clear signals. A doji, in arm 1, for example is usually sign that arm 1 is ending and arm will start next candle. Use range and candle signals from the ATR channel breakout as guides.

Move 1 - starts 3PM (all times my MST local) and can last from 3-10 bars. Even if a move happens on bar 1 or 2 it’s usually better to let it go a bit longer, so consider move 1 min. number of bars to be 3. It is from ~ 3PM - 1AM at most. It is usually a quiet move, but might be the most predictably profitable. You could easily have an EA trade just this arm, moving from quietude to a single big jump that usually happens around 11-12pm. Repeat daily.
Move 2 - starts around 12-1am and lasts until around 4-5am. Can often be the shortest, except when it isn’t, then it can be the longest. Occupies bars 6 - 10 usually, but could go as far as 18. Usually 4 bars.
Move 3 - starting anywhere from 2am-11am, occupying bars 15-21, with the most activity around hours 5-7am. This can often be the longest arm, apart from move 1.
Move 4 - starting 9am and going to the end of the trading day (2-3pm). This is often the least important arm. Ofen you will omit trading the last few hours of the day, and they often don’t form part of the M/W.

Always wait for hourly candle to close.
Always exit or prepare to exit around 3rd std dev.
Exit near time limit for each arm unless in a clear breakout.

Knowledge of this pattern can also probably help you time exits and entries for Reclaimer. For example on an M day it’s probably better to exit a long trade, on the third arm completion, and not wait for close of day.

Update - traded this for 3 days. It’s exhausting. Time-consuming. First day was ok, 2nd day was a write-off because of Trump comments, 3rd day seemed to be going well then completely reversed in the last few hours, adding a 5th leg, (Yellen speech). So this is either valid on ‘normal’ non-news days, or invalid (further checking showed lots of wonky days with no visible pattern, in recent days - wonder why nothing showed up in 2014?)

In addition, knowing which direction price will go, kinda, without knowing when or to what magnitude is maybe limited benefit. File this for now.
Of course, a manual trader might delight in this, and there is probably already some system thread on ForexFactory that basically does exactly this, maybe with the fractal indicator thrown in for good measure. All it is, is the tendency of price to move in Lévy flights, and to revert after moving a certain distance. It's the initiator/generator of fractal chaos though I didn't know anything about any of that at the time of writing, I'm fairly sure.

Halo

I named a lot of systems after video games. Ironically when I started trading I found it to be the most engrossing game of all and I've never really had much time for the hobby ever since. Is this sad? Probably not.

This system spun-off a whole set of similar ideas. The Halo systems. Here is v4 of the eponymous system.

Chart: EURUSD, GBPUSD, USDJPY
Timeframe: D1
Hours: 24H
Indicators: ATR

Entry:

1. Start at the beginning of the trading day. The preceding day’s candle is the signal candle if it meets the criteria in entry rule #4. Enter in the direction of the preceding day. Use high/low of preceding day as the stop-loss.
2. If today was profitable, set stop-loss at 50% of candle height. See exit rule 7.
3. If any candle closes higher or lower than the prior candle’s open (+/- 10 pips*), call this the new signal candle. Enter the following day in the direction of the signal candle. If the signal candle was bullish enter long at market. If it was bearish, enter short.
4. If the candle height is less than 40 points (basically a doji) then ignore this candle for the purpose of signaling. It doesn’t exist. Wait for the next candle. Such candles also do not contribute to exit signals.
5. Stop loss (can be real or virtualto hide it from broker criminals) is placed at the signal candle’s high price for a short order, or its low price for a long order +/- 10 pips.
   1. Optional (true/false): Default true. Move to break-even after 20 pips and jump tighter by 12 pips every 10 pips of profitSee exit rule 7.

6. Optional multi trade (true/false): Once price is 20 pips away from the entry point, place a new order of ¼ lot size lot in the direction of the trend. Move the stop loss 2 pips above entry to protect the profit of the first order. Do this every 10 pips.Consider this for later if the systems seems to work.

Exit:

1. If price breaks higher or lower than the previous day’s high/low (stop loss) level, and if there is time remaining in the candle immediately open a trade in the opposite direction and exit at the candle close.
2. If the order is short AND a candle closes higher than the previous candle’s opening price, exit.
3. If the order is long AND a candle closes lower than the previous candle’s opening price, exit.
4. 120 minutes before close of the week. 10 minutes before
5. After 3 consecutive OR total loss trades, stop trading that symbol for the week. 0 disables this option.
6. Close any profitable trade that retraces more than 50% from the entry point for the daily candle. In other words, act to preserve at least 50% of any daily gain. Hold losses as long as they haven’t hit the stop loss level.
7. If price retraces so that rule #6 is activated AND there are >= 12 hours remaining in the day AND price reverses and returns and then breaks through the prior exit level, immediately open a trade in the prevailing direction and exit at the candle close.
8. Optional: defaults false. Partial close ½ of position if price reaches the 2nd OR 3rd deviation band.

Re-Entry:

1. If there are no open trades, enter at the day’s open in the direction of the last signal candle. Note: this isn’t always the previous candle - it had to have closed higher or lower than a prior candle’s open to be considered the signal.
2. After resuming trade at the beginning of a week, go long after if price has gapped up, and short if it gapped down. (not sure about this one. Maybe ignore gaps?)

Money Management
Lot size - fixed Use small lots compared to account size to handle large swings.
Lot size - percent of balance
RiskPercent
LotsPerDollopofCash
SizeofDollop
UseBalance
UseEquity

Trading Hours
Check broker time, compare to local GMT offset and ensure that the hour count begins on the first hour of the Sydney session.
TradeMonday -defaults true
TradeFriday -defaults true

Safety features
Stop all trades and remove EA if equity closes below SafetyAmount -default -500 units of currency
Stop all trades and remove EA if equity closes below SafetyPercent - default -5 percent of balance

Close all trades when profit exceeds ProfitAmount - default 500 units of currency
Close all trades when profit exceeds ProfitPercent -default 3 percent of balance

MaxSpread
MaxSlip
Send email on any event in this group. If you can use something besides MT4 default email function that would be great since it doesn’t work for anyone.
Send push on any event.
Send alert on any event.
MinMargin - stop all trades and remove EA if margin falls below percent

Flow diagram (outdated) update this for D1 and the new H1 ruleset

This is a nightmare to trade manually so I recommend automating it. I haven't done so yet. The good thing is the rules are clearly stated and mechanical enough that this should be possible. If you do, let me know how it works out.

Covenant

Attached Image (click to enlarge)

This is actually a simplified variant of the first version, believe it or not.
This makes use of the bollinger band 3rd dev. Principle and the tall candle principle (mean reversion)

Zone map:

Attached Image (click to enlarge)

Timeframe - D1
Chart - any
Indicators - 3 bands (1, 2, 3 deviation bollingers)

Entry:

1. Determine dominant trend with centre line of the bollinger bands.
2. Wait for a signal. This is a price movement greater than the ATR(20). The candle doesn’t need to close.
3. Wicks matter. Don’t use them to measure ATR vs candle bodies, but consider large reversals as (potential) signals. They could be supportive or contrarian to the observed direction. The important thing is they show something is happening. Use the candle color (bullish/bearish) at close to determine direction.
4. Wait for a retrace. This could be modest or huge. 20%? But wait for it.
5. Enter when price is unhesitatingly moving back in the expected direction. Not before.
6. Enter in the direction of the signal candle. Set virtual stop loss at high/low of signal candle. If price retraces past that point it’s clearly a defeat of the original thesis.
7. If price doesn’t retrace much or at all, enter once it crosses back past the close of the signal candle.

Exit:
Dominant trend exit:

1. (When applicable) Move the virtual stop-loss to mid-point/centre line when price reaches zone 2/-2. Trail this line.
2. Take partial profits when price breaks into zone 3/-3. Ignore this rule if price started in zone 3/-3 and is expected to wall-climb.
3. Take remaining profits when price breaks through the zone 3 wall. If price started in zone 3/-3 and could start wall-climbing, take partial profits only.
4. Exit trade on a strong (>ATR(20) or use judgment) contrary candle.
5. Exit on hitting virtual stop.
   

Counter-trend exit:

1. When price retraces 33% of the previous move. This is a sign the dominant trend is returning.
2. Take partial profits after clearing one zone. In other words if dominant trend is up, and a strong bearish candle takes price to bottom of zone 2, enter short, and when price reaches bottom of zone 1, take partial profits. If price reaches bottom of zone -1, take remaining profits.
3. Exit when approaching 3rd dev. Line.
4. Exit trade on a strong contrary candle.
5. Exit if price retraces one zone.
6. Basically treat any counter-trend move as suspect and take profits readily until the middle bollinger line shows it is now the dominant trend.
   

Additional Exit rules:

1. Exit when profits are threatened by more than 25%. Hold on to your gains.
2. Exit after 3rd day of trade if there is no/insufficient movement in the desirable direction. This indicates range-bound market and means there was no momentum following the initial burst (signal).

Re-entry:
If price has hit a barrier and is rigidly consolidating, place a straddle, of 1ATR.
AND wait for a strong candle signal?

Gaps & weekend (mostly for trading D1 or higher):

1. If price gaps up or down by more than 25% ignore the previous signal.

Position-sizing:

1. 0.02 when entering
2. +0.01 for a dominant trend trade
3. +0.02 for a trade begun in third zones

Gaps & weekend (mostly for trading D1 or higher):

1. If price gaps up or down by more than 25% ignore the previous signal.
2. Close trades each week?
3. Ignore end of week signals?

Should long wicks up or down count as tall candles (strong movements)? (???) Maybe yes (yes, of course, the wick/body distinction is arbitrary)

Is there a way to identify these signal candles before they finish? Usually start from consolidation. Move quickly.

Can we find a system to measure velocity? Program an indicator or set start/stop zones at a lower timeframe with a time limit.Or just set alerts and if they trigger measure the time taken to arrive at those points. (velocity is less important than you think)

We need to put some emphasis on peaks and troughs. Say if it enters zone 3, exit when it clearly leaves (zone2) ? I think yes.

Protect gains of more than N pips? I think yes. What rule should we make for that?

(at the expense of larger gains?) For example IT MAKES NO SENSE TO TAKE PARTIAL PROFITS AND THEN EXIT THE SECOND TRADE AT LESS PROFIT THAN THE FIRST - so make a rule that says if price retraces to the first exit point, exit at the same profit point. Maybe it makes sense if you are gambling for more.+

Exit trades that don’t produce expected results afer n (3) days? I think yes.

The profitability of this system would be substantially improved if we had a rule for getting in to a signal candle before it closed. (you could use an arbitrary candle height?)

Maybe try setting the ATR (or ATR 1.2) for the day and if it triggers the alert, enter.

This seems perfectly cromulent. It was probably just too tiring to trade manually.

Reclaimer v2

Based on Covenant, but further simplified and making choices based on an indicator made specifically to test the candle height hypothesis.

Use both candle body and range? - trying it. Reclaimer v2.

Timeframe - D1 - tried M15 but it’s just as slow and requires you to be glued to screen.
Trouble with D1 is you need to get up at least once in the night to check trades. (why? Why not just check at open and close of day?) (yes, why?)

Win rate appears to be 3:1 with bigger wins than losses.

Chart - any
Indicators - 3 bands (1, 2, 3 deviation bollingers) set at 20 periods, and ATR channel break-out clemmo 2, set at ATR(5-20), ATR multiplier 1.0, candle ATR multiplier 0.0, no MA filter
I'm really quite annoyed that I can't find the template(s) for this. Did it not survive a computer change?
And use both candle body and candle range for signals,
Uses zone map? Nah.

Entry - Every trade has 4 parts.

1. Signal
2. Pullback }part of the retrace
3. Comeback }part of the retrace
4. Entry point

It's a 'buy-the-dip'!

1. Wait for a signal. I’ve got notification, and emails set up. I might miss ones that arrive while sleeping. Range arrow and body arrow in agreement is a stronger signal than just one or t’other.
2. Check support and resistance at the point where the signal comes in. Check if this is a new signal or part of a series. New signals should get more weight. Is this a red news day? Maybe don’t trade.
3. Wait for a pullback. I think I’m going to make one fast (small) contrary trade after receiving the signal.Exit once the comeback begins. (the origin of the double-agent entry?)
4. Once the comeback has started (how to define it?) Enter in the direction of the comeback.
5. Enter with position sizing according to newness of trend, dominant/countertrend,

Exit - Many factors that get easier to identify with experience

1. Entry was easy, exiting is hard.
2. Exit on stop-loss or take-profit. These are usually set around channel bottoms or signal candle opens, whichever is more liberal.
3. Figure out the hours where action happens. You need to be around I think at 6-7pm, 12:00 midnight, 4am and 7-8am. Demanding schedule, but you don’t have to spend a long time at the screen. Just check conditions of existing trades. Take naps.
4. Look for beginning of the end of the trend - rounding/bending curve, contrary signals, dojis, spiny backs, heads-on-a-pike. Consider exiting on the bend if you have a gain to protect.
5. Exit at or near the 3rd deviation band.
6. Exit with profit to protect at the appearance of a new strong signal, even if it’s confirmatory. (because it's probably a sign of exhaustion) Do the small contrary trade, wait for confirm, re-enter bit. This helps you deal with bounces very effectively.

Number of active trades:
The more trades to monitor the harder it gets.

Position sizing:

1. Add 1 unit for a dominant trend trade. A dominant trend is defined by the midline of the bollinger bands or a higher timeframe
2. Add 2 for a wall-climbing possibility
3. Add 1 for a signal with 2 arrows in agreement
4. Add 1 for a signal that begins a trend
5. Deduct 1 for a signal with 2 heads - maybe don’t trade it.
6. Deduct one for a counter-trend trade?
7. Reduce position size steadily as a trend matures.

This idea of adding/removing volume based on the number of factors in your favour is interesting (Donnelly-esque) and I never came back to it. Why? Probably because with small accounts, reducing volume is the main consideration.

Support and Resistance

1. Watch for major levels apparent on W1.
2. Be wary of trades that would take you lower or higher than these levels.
3. Deduct 1 position size unit for trades that would have to break through support or resistance levels to succeed. Or don’t trade these.

How often do you get a signal that doesn’t retrace at all? Need to check but initial study shows ...rarely. More common in later stages of streaks, after the first signal has passed. Update 8/23 signals since September on USDJPY D1. So about ⅓ of signals will not pullback immediately. Only 2 of those 8 (¼) didn’t have a pullback the following day. So only 2/23 or 9%. (there's that number again) Unscientific, but worth checking on other charts.

And in most of those cases the pullback happens immediately the following day. In most cases though, it isn't’ worth waiting 2 days. Just expect the momentum to not last long without that prompt pullback. Is it better simply to avoid these trades? No? (it likely is better)

Can we quantify when something has pulled back and when something hasn’t? Of course.
Can we quantify when something has come back and when it hasn’t? Of course. The key is to have a time horizon.
Can we identify the most likely time of day for the peak of the pullback? The comeback? Yes, to the first, the second, I'm not sure.
Can we identify the most likely percentage of an initial move that the pullback will take? Statistically, yes, but it may not do much good for forecasting when to enter.
We need to start gathering data. Always. Gather data and test. That's what you do.

Gold
Gold is much ‘streakier’ where the exception is a next-day pullback. ⅔ don’t pullback the next day(?) Sometimes it will happen a couple or 3 days later. There doesn’t seem to be a point in waiting/trading a pullback or waiting for comeback. In fact, when the following day trade is contrary this is more often a signal of the end of the trend. Let’s check. We need some way to avoid dead cat bounces though...do we? **

Update 2 - I may be wrong (you often are) - look at all the reversals starting around Q1 2016. Unless gold is ‘streaking’ it seems to act just like currency in this respect. In fact reversals are the only thing worth trading when it’s ranging. So the question becomes - how to define when gold is ranging and not trending? Now you're getting it.
One way - when it’s on a big trend it’s climbing the wall of the 2nd/3rd deviation band. If the reversal doesn’t happen next day it’s a signal that it ain’t happening. Or just delayed. Faking.
Reversal seems to end up being same height as signal candle. (???)
Double signal seems to be more genuine motive force than just a single, but that’s not guaranteed.
Smaller ‘Stair-case steps’ - seem to be first sign of a genuine longer trend.
Avoid trading on Friday. Gaps can be big over the weekend. (or just close end of day, or just trade smaller)
Heads on a pike seem to matter less with gold. (now I remember - this isn't a candlestick pattern, it's a breakout indicator pattern. Shame I didn't screenshot it...OR DID I?)
2016 - if I disregard pullbacks - 32 wins to 20 losses. About ⅔ wins. Pullbacks common in ranges, uncommon in trends. A pullback is a sign of a range. Naturally.
The only question remaining in my mind is how to quantify when you have started the comeback. Need to wait for it to go ahead of the pullback?
A very large pullback (70%+ of the signal) is a sign the original signal is very suspect. (or is it? These days that's normal. 80%? Time to worry.)
** Update: Yes, wait for pullback, and wait for comeback, just as in currencies. If pullback doesn’t happen, enter as usual. If comeback doesn’t happen, stay in the reversal mini-trade.

Double-headed candle with a huge range seems to be an indicator of a shake-up. Trend is reversing or volatility is increasing.

Questions:
Use trade manager - like Gecko’s? To handle exits? Automate the whole kaboodle.

What can we say about the ATR(20) when it’s at a local historic low? Do things improve in terms of predictability?

Is it worthwhile to do a stock-bee style consolidation search and stalk trades from that point? (years later and I'm still pondering this. Just try it already.)
Place buy/sell stop just outside the ATR to avoid ranging nonsense.
Checklist ideas:

Before you start:
-Backups running
-internet working
-2 calendars up (don’t rely on myfxbook or fxstreet alone!)
- some news read
-What session or time period are you trading? _

Before placing:
- Max number of trades NOT exceeded? (if so, close some down)
-Is this a spontaneous seat of your pants trade, or a trade according to a system you’ve devised?

1. Near 3rd std dev band? (stronger reversal chance)
2. First trade of its type (up/down) in sequence? (weight first trade more than others)
3. Size of previous retrace? (useful?)
4. Signal: range arrow, body arrow or both?
5. Doji before or after signal? (signifies what in each case?) (turnaround, start of range)
6. Head-on-a-pike signal (or before/after?) - signifies increased chance of reversal
7. Currently in range/trend - mature or new?
8. Currently wall-climbing? 2nd or 3rd band?
9. Presence of:
   1. stair-case steps? (break out possibility)
   2. Double-headed candle? (reversal possibility)
   3. Strong historic support/resistance level?
   4. A gap? (it will fill in)

   
   

10. Is it:
    1. Friday? (caution)
    2. A red news day for the entire market? (caution or avoid)
    3. A red news day for the pair?
    4. Extremely consolidated (series of very skinny ranges, on a horizontal plane)

    
    

11. Exit plan:
    1. ATR channel
    2. Support/Resistance level
    3. Trailing stop
    4. Trendline stop
    5. Others?
    6. _______?

Before quitting:

1. Stops and TPs in place?

When to move stops?

1. To breakeven after 1ATR gain (that’s it)

LOL! With great difficulty. I am constantly tinkering and messing things up to my own deteriment. I thought automation would fix this but it's clearly a flaw in my personality. That's one reason I post here so often; at least it keeps me busy and slows down my 'interference rate'.