Friday, May 31, 2013

Order flow and its impact on chart highs and lows

Last night as I drifted into sleep, my mind churned over order flow and I developed an idea. How do you tell when a market is turning?
 
Price action traders would tell you to look for price reversal patterns, but in my backtests, these signals alone do not provide a tradable edge. While these patterns do suggest a short-term reversal, that's all they are, short-term.
 
Trading these signals at significant support and resistance levels may add some edge, but many times they don't work. Could I add something more to add to this edge?
 
I thought about order flow, and how the market can show its hand on the charts. I thought about the actual pattern of a market reversal. If we have a bullish trend, we normally see higher highs and higher lows. We only a see a reversal when this pattern is broken and we see a lower high.
 
Why do we see a lower high? In terms of order flow, we are now seeing an increase in sell orders and/or a decrease in buy orders. Buy orders are being exhausted. But more importantly, they are being exhausted earlier. This is how we know that the market has a higher likelihood of turning. And I suspect this likelihood increases on the higher timeframes since it takes a lot more money to move the market, reducing the possibility of manipulation.
 
Here's an example of what I'm talking about. It's the AUDSGD on the daily timeframe and we see that price has hit a support level. But we don't know what will happen next. Will it break through support and continue to fall, or will it reverse?



If we want to trade a reversal, we can wait for a price action entry signal. But, if we are to trade with order flow, we'd want this entry signal to signify a higher low. This tells us that shorts are being exhausted earlier, signalling a growing glut in longs.


Intuitively, I think this can add an extra edge. Trading reversals is always risky since you're going against the longer-term trend. However, I feel this is a good way of filtering an entry.

Thursday, May 23, 2013

Implied volatility and order flow

I'm now beginning to mix some fundamental analysis into my trading methodology. I guess you can say that I'm evolving. :)
 
I performed a historical volatility analysis on the majors and popular crosses. Each pair does possess a personality that usually reflect their economic and geographical relationship.
 
As part of my studies into order flow, I've been examining ranging markets and how they behave. If you are going to delve into order flow trading, I think ranging markets would be the easiest to get into. Demand and supply zones are quite easy to mark, as well as your stop loss and profit target.
 
My historical analysis involved finding the highs and lows of each pair from 2001 onwards, and dividing the high by the low. This would imply the historical volatility of each pair.

The results:


An alternative way of analysing historical volatility would be to simply whip up the monthly charts and plot ADX over a large range. In this case, I chose a period of 120 months (which equals 10 years).

The results:

 
By comparing the results from the two analysis, we can see that the least volatile pairs are the AUDNZD, AUDCAD, AUDCHF, EURCAD and GBPUSD. After doing some research over the internet, the EURGBP and GBPCHF are routinely mentioned as good pairs to trade ranges. From my analysis, though, they don't seem to be the "best".
 
If you're going to trade ranges, the AUDNZD would seem to be your best bet. Australia and NZ are both commodity currencies, both are economically developed with similar exports and trading partners, both are culturally and politically similar, and both are in close geographical proximity.
 
What I found surprising was the relative lack of volatility in the GBPUSD, being a major pair. If you want to range-trade with the least cost, I think the GBPUSD would be a good choice.
 
What does this have to do with order flow?
 
My analysis attempted to profile each of the majors and crosses and figure out their "personality", whether they tend to range or trend. The behaviour of each pair is dictated by big money. Us retail traders are much too small to have any influence. However, by getting a snapshot of how the big money traditionally treat each pair, it's almost as if we're taking a peek into their order books. In a range-friendly pair, we know that the big money are more likely to reinforce support and resistance within a narrow range. Trying to trade breakouts would be more dangerous. It's all about finding and accumulating edges.

Friday, May 17, 2013

Still alive...

I'm still trading, but have shelved almost all of my trading systems until the end of June 2013 (end of the Australian financial year).
 
I'll summarise my activities for the last few weeks.
 
Market microstructure and order flow
 
If you listen to most traders, even experienced ones, they'll say that the 1M and 5M charts are mainly noise and should be ignored. This is not true. In fact, there is some startling structure that I've witnessed and this discovery has helped my understanding of the market immensely.
 
Here's a gold 1M chart from today. It's a naked chart. At first glance, you can't make head or tail of it. It looks like the market is moving erratically.
 
 
However, if you take the time to really look at it, you'll begin to notice some structure in the chart. Specifically, you'll see a period of ranging prices before a breakout. I'll mark the ranging periods below.
 
 
It's not precise, but generally once price breaks out of each marked box, you see a large spike in volatility before price re-settles into a new range.
 
Does this look like noise and chaos? Not really. In fact, you could probably trade these ranges. (I did a quick backtest, and it's enormously profitable if you ignore the spread. Sadly the spread kills it). 
 
Now, why would price range for awhile before breaking out? It has to do with order flow. I've been aware of order flow for many months now but didn't pay too much attention to it. But whenever you see price ranging for a period of time, that's telling you the big money is trying to accumulate positions without moving the market against it. For example, if the big money is trying to accumulate longs, its actions may move price up before it has accumulated enough positions. This is bad. Their main tactic, therefore, is to accumulate longs over a period of time. If price begins to move up, they slow down their accumulation until price starts moving down again, at which point they resume their accumulation. This is why you see a ranging period. The big money is pushing down and then easing back on the accelerator.
 
At some point, the big money's accumulation will exhaust all sell orders within that price range. Liquidity has dried up. It's at this point that you'll see price break out of the range since anyone who wants to buy, must now bid higher in order to get something.
 
This is just a simple, badly-explained example of order flow.
 
How does this help my trading? Well, you can see those ranging periods of accumulation when you zoom out to higher timeframes and see small candles. Generally, a small candle will be followed by a larger one.
 
Most of my effort for the last week has been concentrated on designing a trading system exploiting this fact. My Hermes low volatility system is based on this logic, but I'm trying to figure out a way to trade accumulation / distribution more consistently. The trouble with my Hermes system is that you only get a few trades per month, but accumulation / distribution happens ALL THE TIME. This is one of the cornerstones of market behaviour.
 
Anyway. that's what I've learnt over the last few weeks. I hope to post some discoveries in a future post.
 
Van Tharp Trader Test
 
Last week I also undertook the Van Tharp Trader Test. The test attempts to classify your trader personality. For example, you may be a Spontaneous Trader, or an Administrative Trader.
 
It's a simple test, only 35 questions long. My result was Strategic Trader, one of the two "best" personality types. I'm taking the result with a bit of salt, but am curious what other people got? It only takes five minutes to complete.

Tuesday, May 7, 2013

Trade Week in review: 29 April to 5 May 2013

Last week was pretty bad. I won a few trades, but at one stage I had six trades open at 1% risk per trade. Five of those trades began to move against me and I was down -3%. Rather than wait for my stop losses to get hit and lose 5%, I decided to close all trades and take a break from trading.
 
My main goal for the next month or two will be to revisit my systems and see if I can tighten my stops and create reasonable R:R ratios. At the moment, most of my weekly systems use a reward-to-risk of 0.25:1 or less. While it looks profitable on paper, these sorts of ratios present a psychological risk in that when you encounter a loss, that loss really hurts. It would take a relatively large number of wins to claw your way back. Now, when I was looking at the prospect of five losing trades, the psychological weakness of a low reward-to-risk system became really apparent.
 
The other problem is that because losses are relatively rare, you can't really build a psychological immunity to losing. In contrast, say you have a system that wins 50% of the time, and loses 50%. You can expect a loss when you enter a trade. It doesn't bother you so much because it's well within your system's expectation. On the otherhand, if you have a system that wins 90% and you experience a loss, it puts you on pause. Suffer a few more losses, it makes you question your trading.
 
Trade management is also becoming an issue. Having five trades move against me at the same time is the result of short-term correlation.
 
These are new trade management rules that I am contemplating to minimise correlation:
 
1. Only have one trade open PER TIMEFRAME e.g. I can have one trade open using a daily timeframe system, and another trade on the weekly timeframe, but I cannot have two daily or two weekly trades open simultaneously.
 
2. Tighten my stops and create reasonable R:R ratios. Tighter stops means a shorter duration for trades, and thus less overlap and correlation. This will also allow a higher frequency of trades.
 
I will be going on a "semi-break" for the next few weeks at least. I'll continue trading with one trading system that I feel is robust enough (my Hermes low volatility breakout on the daily chart). I'll be re-optimising everything else with tighter stops. This may take awhile. :)