Some MFE MAE findings

Mostly writing this one just to have record of what I worked on here since I generally don’t keep all the plots I make when doing analysis.

One of the things I’ve really been working on and thinking about is making as much of my analysis as possible in an extra dimension – that is, from a simple frequency distribution to a scatter plot.

I created a simple strategy of sorts (more of an extrapolation of a model with some filtered parameters) and in general just wanted to learn more about it. Putting the strategy into a trading simulator, pulling out all the data I want to examine, and plotting it is a good way to do this.
First:

This is simple win/loss in green/red with risk/reward ratio of the y axis and hours til trade completion on the x axis. Thus you can see already by this point there has been some filtering of the system (no RR greater than 10, no trade duration greater than 30). These other points not pictured can either be removed completely or analyzed separately.
Second:

The next iteration takes the data and looks at MAE/MFE, with winners in green and losers in red. More filtering and improvements can be made here: almost no winners have MAE lower than -.01 (100 pip drawdown) and no losers have MFE greater than +.005 (50 pips “left on the table”). Below is a more blown up image of the negative region:

Thus it’s an easy adjustment to say if a trade reaches -50 pips in drawdown, it’s probably easier to just cut it as a loser and save additional risk.

Lastly:

Trimming off the +50 and -50 gives a better look at what the rest of the trades look like. Then if possible, take a look at the new scatter plot, made observations, and go through the process again; Making trade adjustments, looking through individual trades for patterns, or moving on to the next project.

Advertisements

First Look at MTF analysis on transient waves

Model: Fractal n=24, Transient waves

Here’s my first look at MTF. I took the model I had for the 1 hour, and simply multiplied it by 4 and applied it to get the equivalent 4-hour chart. I was curious about the ideas of traditional wave theory; That smaller waves have a 5, or 7, or something legs that are equal to 1 leg of the higher frame. I don’t know much traditional wave theory, and I don’t expect my model to be that perfect instantly, but I was curious about to just explore: what kinds of 1-hour legs are contained in 4-hour legs?

A simple “T” is pictured on the right – A single trend pattern (3 legs). A 1, in this case, means that the leg completed in 1 move – the 1-hour leg and the 4-hour leg are identical. A single leg, simple ABC wave, a 2x ABC, and 3x ABC leg make up an overwhelming 80% of the wave formations in 4 hour frames.

This is somewhat interesting because while present, it shows that TEE is actually a very rare type of way to complete a leg.

It means that in the scenario above, as price is rising on the 4th leg, the soft resistance is a marker for how price can react. One way to look at it is to say a short at this level is potentially a good opportunity – At this point, it is not a “1” move. It is either a T move, and the move is over, or it is more (perhaps TST or TSTST). It is fairly close to 50-50 here, so the risk:reward here is crazy (Now, what is the probability that a move makes it this far and continues down? That’s another test). However, if price were to break into the “unlikely” area, now we are basically in reversal mode. It would be a good chance to load longs by the blue box because breaking the lows here would be very unlikely.

It’s not perfect because sometimes a full TST wave will have already been created on the 1h frame before the 4hr frame is “confirmed”. It’s easier to trade when the 1hr and 4hr frames are confirmed in the same leg, and even then it’s not always a sure thing. The need to “backfill” full waves occasionally on this model is it’s biggest downside.

completed:
-Learning the Oanda API to create an automated trading bot. Learning MQL seems a bit much for a programming novice like me, so if I can do it in python, why not try?
-Learning how to draw lines with the matplotlib library to better create pngs with wave lines drawn in
-Additional work on the main framework to allow testing new ideas/models easier/faster
-Trying MultiTimeFrame analysis
-Automate data collection into sql database (woo!)
-Begin test/live deploy of first automated trading bot

to do:
-Work on new model

It’s Already July

Wow! A small self-blog and an update. I had to scrape my initial plan for the year because, after some research, I realized I didn’t quite have the tools (yet) to be able to play with. Maybe in another couple years 😉

It’s been a slow transition, but I’ve finally made it to the point where my self-taught skills in python are about on par with where my excel skills are. For quick things, I still default to excel, but programming is so powerful. It should absolutely rank #1 on things I regret about my trading career. Rel suggested it to me one of the first times we spoke, and I didn’t follow up on it. I was too eager to get as much done ASAP, and couldn’t be bothered to take the steep learning curve to learn how to do things the “right” way. I did learn a lot about excel, but most of those skills are useless now. It’s a shame really, but I did eventually learn. Programming, if it’s not obvious by now, allows iteration SO much faster. It allows one to port a model into a higher time frame, or to be tweaked and run again almost instantly. It allows you to run “what-if” scenarios and trade simulations much easier and can help you visualize so much better as well. I can create a way to model waves, iterate through data, create a OHLC charts, and save them into PNGs to view individually and think of new ideas. Again, Wow. I’m still learning a lot and simple things to explain take me days/weeks to complete, but I’m getting there and it’s worth it. I’m much busier these days so I don’t have nearly the number of hours to spend on trading related things as I used to but the dream will never die. Back to work.

Things I’m working on:
-Learning the Oanda API to create an automated trading bot. Learning MQL seems a bit much for a programming novice like me, so if I can do it in python, why not try?
-Learning how to draw lines with the matplotlib library to better create pngs with wave lines drawn in
-Additional work on the main framework to allow testing new ideas/models easier/faster
-Trying MultiTimeFrame analysis

The New Years Resolution

Return to research. I’ll cross these off as they occur 🙂

1. Learn the tools (estimated time 7-9 months)
2. Complete initial research (1 month)
3. Tweek and adjust (1-3 months)
4. Complete write up (2 weeks-1 month)

Like most research, it will either reveal a large edge in the market, or be completely useless. However, either way, the write up should provide a great summary of everything that has been done on this blog and something that everyone can understand. I’m excited! Tentative finish date is, of course, the end of the year. Let’s get to work.

My Final Post.. for now

If you can’t tell from the lack of posts I’ve been making, I haven’t done research in a while. And for the first time since I started this blog, and for the first time since I started my journey as an aspiring trader, I don’t plan to do any research for the foreseeable future. I want to structure this final post in 3 parts: The personal side, the analytical side, and the future.

On the personal side, having to quit sucks. It really sucks. Having to admit to all the people who told me I would never make it that they were right hurts. Having to invest literally thousands of hours (I would estimate 4500-5000) and get practically nothing out of it is depressing. Trading for a career was THE career as I saw it years ago – full time, hours a day watching the charts like Morse code, seeing things 99%+ of people couldn’t see, acting on part gut, part knowledge, seeking the dream of being an elite at something. I could go on and on, but the short answer is it didn’t happen. Perhaps the focus should really just be on one word: why?

Per usual I’d like to think that there were 3 main reasons:

First: Lack of useful screen time

I’m big on visual learning and learn by doing, and trading is no different. I’ve learned over the years (in a lot of aspects of life) that to learn things well, I take them and break them apart into as many different parts as possible and analyze them to death until they make sense to me. Seeing the actual process of getting into and out of a trade was something that I never quite got the hang of. Sure, testing in the lab I got a good sense of what structure I want to trade being given the candles, but I’d argue that there’s a difference between trading OHLC candles in and of themselves (really only trading the open and close) and trading the candles AS they are forming. Getting an idea of what “live flow” looks like isn’t something I’m familiar with. I felt that not being able to actively trade the London and NY sessions hurt my ability to learn this feature of the market greatly. Additionally, not being able to piggyback another person’s trades (understandable) or viewing someone’s past trades (also understandable) added on to this. Let me explain. In so many other trades (pun not intended) or elite skills, such as poker, chess, strategy games in general, there are replay features, streams, etc. It allows a user to passively absorb the “correct” trading environment. Junior pit traders and the like get mentored in this aspect, and it was one I think would have been very helpful to me.

Second: inability to network effectively

This is an area that while I wish it could have happened; I think it was simply a goal and not an expectation. Many people make it without it. To be specific, I was unable to find another person, or group of persons, to share ideas with on a peer-to-peer level. I’ve found in my numerous encounters with people that as aspiring traders get closer and closer to a working system, they guard it closer and closer. New traders happily share all their ideas about how they think the market works (unfortunately generally disproved statistics) but more seasoned traders either a) speak very vaguely about how they trade or b) trade a specific way or have a certain understanding of the market and are adamant about not straying from it. Again, I don’t wish that I could have someone spill all their secrets to me, but having someone who could both test their own theories and challenge/offer input in my theories would have been the ultimate “2 heads are better than 1” situation. I met a lot of “thanks but I think this instead” which isn’t very useful given the scope of the availability of forums already. Not a requirement, but another “nice to have”. I’ve shared all the relevant findings in my career, but I don’t expect the same of everyone else.

Lastly: Inefficient use of time and incorrect initial analysis.

I think this might be my biggest reason, and this one is my fault completely. I think I picked the wrong approach to the market. I spent so much time analyzing waves and movements in micro frames (30m/1hr) and not enough on the big frames (d1/w1). Looking at micro frames absolutely requires one to trade full time to effectively make use of it. As stated with screen time, I never had time to actually trade what I was studying. IF I found a big enough edge, I would have had to go straight into full time trading and hope that it was enough to make a career out of it. Wrong wrong wrong. So wrong. In hindsight, my thought process was simply: have no knowledge->acquire edge to trade full time ->trade full time. However, I should have taken an extra step or three and done something more like the following:

have no knowledge->acquire edge to swing /position trade(1-4 trades a month) ->earn some extra income (5-10k/year)-> continue to acquire knowledge to swing trade smaller frame (1-6 trades a week) -> earn substantial income(10k+ a year) ->gain enough knowledge to trade intraday = trading full time.

Aren’t time frames just reflexive of themselves and they’re all the same? No, not really. Trading on a larger frame grants a lot more wiggle room to be wrong (since position sizes are so small). You can be saved in a position trade in so many ways. Unexpected news, expected news, interest rates (carry trading), light weight martingale strategies, random chance, etc. The probability that a specific point will be revisited is obviously much larger when you allow weeks rather than hours/days. You can afford to miss your entry by 50, 60, 100 pips and it’s not that big of a deal. In a live environment, it’s different.

When I think about scalp and intraday traders versus swing and position traders, I think of this comparison: precision vs. prediction. I think scalpers are extremely niche in their knowledge. They know a few things and they know them damn well. They’re reactive and quick. Swing traders on the other hand are more about preparation. They have to continually adjust their outlooks based on news, and rather than react, they anticipate. That’s how I see it anyway.

If I could offer one last point as to why I failed to go fulltime, I’d simply say that I was missing “it”. Not seeing the things that other people couldn’t see.  Not being creative enough. Not thinking outside the box enough. Not thinking inside the box enough. Not enough coding power. Not enough preperation. In my opinion of trading, the ends does justify the means. If you failed, you have no one to blame but yourself. There’s no rigging. There’s no luck. It simply is. While I still have thoughts that I could do it one day, I couldn’t do it in the prescribed timeline.

And with that, I’m off for now. If you were looking for the happy ending, so was I. It’s not all doom and gloom however. I have learned a LOT in the past 3-4 years. I’ve surprised myself with how well I can work data in excel, how easily I can manipulate raw data into swings for analysis. I could probably be more efficient in anotehr language like R, but perhaps that’s for another time. A reboot of sorts. I know so many FACTS about the fx market. Not opinions, not thoughts, not theories about the “market marker” that is out to get us. Statistical, raw, empirical evidence about how price moves. And that’s kind of cool. Maybe I do have the edge. Maybe I just need to work it over time and play my spots that I know. There’s always tomorrow, another trade, and I WILL trade.

With that, I’ll be on vacation until the end of the month, and hopefully by the time I get back I can upload one last live account to myfxbook. Until then, remember: Question, Research, Profit.

-LiquidGenius

RHS # 2.9 Ratios separated by type

one line for down waves, 1 line for up waves, filtered by smaller ratio. A nice idea, which is why I’m posting it, but atm i think it might be a bit too choppy to trade with.

Market Summary 2015

Here’s a bit of a summary of where I’m at and what I’ve been thinking about lately, to help me re-focus where I should be putting in the hours for the next year.
Market theory:

1. The market is more or less most efficient when viewed in the 30m 7bar wave length (“The” chart)
2. Retracements are much more common than rejections or trends.
3. Markets do not often contract continuously (retrace->retrace->retrace) nor continually expand (trend->trend->trend). Happens, but not often.
4. It is much easier to predict where a retracement will end as opposed to where a trend will end. This is pretty easily backed by theory that retracements are occurring in previously interacted zones (S/R) as opposed to trends that may be operating in “air” space where market participants are unsure of what the correct reference point is.
5. Retracement moves occur with a minimum of 25-30% of the previous move. This may be have something to do with “minimum” profit taking at certain levels.
6. The 40-80% range is where most retracements end. There are many “seen” levels at this point (50% half point, 61.8 fib), so it will give reason for a lot of people to jump in. If it’s enough, price will hold in this range.
7. The 80-100% is the “danger” zone. If the level is being protected, it will usually happen here in the form of a large wick. The highs/lows may occur in this region, but closes (I think) are likely to be somewhere in the 60-80% range if it holds.
   -I think candles who’s high pops just a little over 100% but close under tend to then make a 30% retracement (need to check)
8. 100%-120% is sort of a limbo between a breakout and a major rejection zone. Prices that reject this level tend to be similar to the level before this, often forming large wicks before shooting the other way.
   -It seems that the 80-120% range is where price is most unsure. I don’t really know of a good way to measure this, but it might be worth clearing up how price moves in these areas.

 

Following from the above, here’s what I’ve been thinking about in terms of how to apply this.
Trading Strategy Theory:

1. Since price is limited in what it can do within the soft zone of 40%-80%, it makes the most sense to try to pinpoint a price within this area and trade it.
   -Doing so will set a hard stop at the break of the level, setting a max risk point.
2. Pinpointing a single bar within the range will help minimize risk further and create a more ideal risk: reward ratio. This will reduce accuracy and result in taking multiple small losses, but open up the potential to ride out trend moves when they occur.
   -There are two methods to do this: either predicting that the current level will hold or waiting for the completion of a signal (3 bar fractal).
   -Alternatively, one can start small and continue to build positions behind (averaging losses) and maintain the “max risk point” from above. Doing so will allow more opportunity to get out of failed trades with minimal losses, but will reduce profit in trades that work out on the first attempt.
   -Trading with an absolute stop (at the 100% break) would require holding trades longer than what is normally the minimum expectation. This basically creates a fixed risk, variable reward. This may not be desirable because one wants to play as close to the “safe numbers” as possible and as frequently as possible.
   -A possible way to improve profits in this area would be to then stack trades in favor of the move (averaging up)
   -Since there is a roughly predictable number of bars that can occur within the soft zone, filtering out even half of them should allow for a very finite, crisp window of opportunity.Doing so will eliminate trades that do not occur within the soft zone (where no signal is possible, <30% retracement), as well as trades that do not provide a signal, while taking failed trades (where a signal appears but price continues to reverse past the 100% level). Trading this kind of style would require a lot of patience to stomach losing trades, since one would lose out on trading 2 potential winning areas while maintaining the losing area (since it is unknown), as well as the necessity to capture the winning trades to make up for losses and then some.
   -One must be very careful when filtering out trades in the soft zone. Since retracement moves are the most common, inaccuracy will lead to losing out on what is a likely opportunity.
3. Since trend ends are much more difficult to anticipate, it is probably best to play into them rather than out of them. That is, hold trades and get out when a warning sign appears, rather than getting into a reversal trade at that point.

Statistical backings:
Currently I am using two types of edges.

1. Histogram and curve distributions – taking all the data available, and matching the points of interest against them. If 90% of interest points occur between the values of 0 and 1, then if the current point is not in that range, it is likely not a point of interest.
2. Max counts – Counting the frequency that a specific pattern or “signal” will appear within a set (range, wave, etc). If 90% of waves contain 3 or less patterns, then betting once the count is 2 or 3 is a good spot to be in.

My favorite histogram stat might be this one. The fill bars are okay, but the frequency of these bars isn’t high enough in my opinion to rely on them as an active trading tool. The left side range reference isn’t the strongest, but it is clean and simple to understand. My favorite count stat is probably this one from FF. The skew is there, the timing is implied, and the occurrence is decent. Trading after the first one shows means losing out on 30% of the trades, and there is a remaining ~60-65% edge.

It may be worth trying the above two figures out and seeing what issues I come up with, since my work in the rabbit hole is, well, a deep deep journey. If I try to follow the little pieces of gold from the more experienced, perhaps my aim should just be to continue to reduce risk and just sit on winners. It might be better suited for my lifestyle too.