Your question was a puzzler to me too, we certainly couldn’t have set 22 and 12 as extremes if we traded 10 years ago. The more urgent question to me now is whether the 22 and 12 work today hahaha. The 200 MA rule
One other thing is that it reflects the price of gold itself. In other words, if that period saw all currencies during that period had a ratio lower than 12, then the price of gold itself was undervalued and we should buy gold. The problem is we are looking at data of the 90s and early 2000s without context, we have know idea what the fundamental situation was back then so of course we will not be able to explain what happened.
I think for now its better to look at how price reacted when we closed below 12 or above 22. Or closed below and above the 200 Daily or weekly MA.
as I see, I took the weekend of with Rookie, as there are not many things in the COT report. You four decided to discuss an interesting topic. I really like it. It looks interesting. Philip said we should ask questions or raise concerns if we have any.
Here are my two questions:
Why do we suppose to use the 22 and 12 values for all currencies? Only because it worked on one instrument it does not confirm that the values are true for all pairs. I see that Philip wrote something similar in his last post.
Even more important is the chart in one of the articles Peter linked. Only because the gold/oil ratio is very high, and the ratio drops it doesn’t mean that oil will rise. We only looked at it from this perspective. Keep it in mind, if gold drops and oil is still cheap, the ratio will still drop. This is explained great in the article as I said.
I agree with you on point number 1 completely. I just used 22 and 12 as a starting point since they were the numbers used on oil. Back testing is what will help us reach a number that represents extremes. BB shared a brilliant idea of using bollinger bands as extremes. I’ll look at it.
As for point 2, I understand what you are getting at. However from a purely mathematical point of view, if gold drops and oil remains cheap, the ratio will actually increase.
I will read Mike’s recommended books to see how the author dealt with these ideas.
Its also important to note that the foal from polishing the ratio is not to trade. But rather to help us with our bias. Trading will still be done using COT report and our technical system
Having the ratio would help me for example on a pair like GBPCAD or AUDNZD. I do not know what the trend is on these pairs. But if we do develop that ratio it might help us reach a conclusion on where the pair stands.
I asked my broker if my account can go into negative balance or not. Here is the answer:
“In regards to your inquiry, your statement is correct and your negative balance is legit. Please be advised that you are responsible for settling any deficit balances in your account.”
How about you guys? Have you asked about this? Does someone has a reliable broker where he knows that balances cannot go into negative amount?
I do not know as it hasn’t happened to me yet. I know that there are a few who do that but the absolute majority of brokers have that negative balance. Hence the warning that you can lose more than you invested. Sorry I couldn’t be of much help there.
I also do not know it works in practice as I have also not experienced it. The reason why it is important is because of our discussion yesterday. If the account can infact go into a negative balance then our whole money management discussion brings nothing in the end because we have to settle the losses in a later date.
No, first you get the market call before you get into negative territory. Second if it goes to negative it is up to the broker to chase you to get their money. You won’t pay it that easily, especially if you’re not happy with some of the execution of stop losses etc.
So I read the chapters Mike recommended in his book. They are good, however what the author discussed has no predictive value. It is a similar method to Mike’s currency ratings and I like Mike’s more.
I also tried Bollinger bands but the problem with them is that whenever volatility increase the bands will follow, so its hard to mark extremes.
I also read Sure thing commodity training by Larry Williams. He has an indicator called Will Vall that does exactly what I’m trying to do. However he said that gold should be accompanied with similar ratios but with Bonds and Dollar index. I have not seen Will Vall in action and so I’m reluctant to try and use it especially that it is much more complicated.
One idea I’m trying now is to use William %R to determine extremes. Whenever the ratio gets overbought, then a rally is near by in the instrument. Whenever the ratio is oversold, then a sell-off in on the cards.
This should solve the problem BB discussed that at many times price is below or above the extreme for years. It also addresses FE’s concern that just because the numbers work with the gold/oil ratio they shouldn’t work with all other instruments.
I’m in the process of testing it to see if it works and will let you know my results.
Ok a few pointers: 1) Overbought and oversold levels are 5 and 95 respectively.
2) For the ratio to be considered overbought, it has to cross the extremes by at least a full point. I will not consider things such as 95.02 as oversold. This is done for the sake of simplicity as I dont want to share a picture that has a million signals on it. We are still testing if the ratio has any use, we don’t know that yet.
3) I have done the testing on the monthly chart, this is because it will very difficult for me to back test years of data for every currency pair on the weekly or daily chart.
4) Theblack line is the ratio, the red line is AUDUSD.
First vertical line from the left (February 2008, AUDUSD price: 0.93. %R overbought, AUDUSD has or is about to bottom.) What happened? Price dropped the following month to 0.91 before entering a 3 month rally to reach 0.96. It is worth noting the ratio failed to warn us of the fall in price that resulted from the 2008 economic crash.
Second line: January 2009, AUDUSD price: 0.64. %R overbought, AUDUSD has or is about to bottom.
What happened? AUDUSD entered a massive, multi-year rally the following month. It reached as high as 1.1 in April 2011.
Third line: September 2012, AUDUSD price: 1.04. %R overbought, AUDUSD should be about to rally.
What happened? AUDUSD went sideways for a number of months before entering a major sell-off in March 2014. Again, the system failed to warn us of the impending fall. %R came very close to an oversold level just one month before the sell-off, however the 92 reading it produced is not considered a signal of a top forming in AUDUSD. It is the second time the system fails to warn of a sell off (more on that in the outcomes section.)
Fourth line: January 2015, could AUDUSD about to rally?
Outcomes
This test gave some evidence that the ratio can be used to spot tops or bottoms in a forex pair.
However, its failure to warn of some reversals convinces me that his tool should be used as an aide to our main tools of COT report, technical analysis and money management.
The system failed to produce a single oversold level in the tested period. This is more a problem with the %R than with the ratio. It is well documented that the %R is a biased indicator (so is the stochastic): It is more likely to be more accurate in one extreme over the other, this is usually used by technical analysts to determine the trend of the pair. Based on that, we say that the AUDUSD chart above has a bullish trend because the %R failed to record oversold levels.
Work is needed to overcome the drawback in %R, one way that would solve that is may be using smaller time frames (such as the weekly) in tandem with the monthly chart. Your ideas are also welcome on how we can overcome this issue.
More tests will be needed before we can reach a proper conclusion on the ratio.
It is also worth noting that I looked at a monthly chart of AUDUSD with the same %R and noticed that it gave completely different signal. That comparison showed me that on this particular test, using %R of the ratio gave an edge in predicting bottoms. Using AUDUSD alone was basically useless.
I’ve got a question. I’m trying to get familiar with this ratio thing, I even started reading Ashraf’s book but since I’m working, the progress with it will probably be slow.
First, let’s get some things clear.
The ratio of Gold/AUD for example means, that for one ounce of gold, I can buy X (the ratio’s value) amount of AUD, right? So, if the ratio is rising, Gold is getting stronger relative to AUD.
Actually, I’m pretty excited about it. Now it’s the perfect time to take a step back, get my mind off the issue and come back to it tomorrow with a clear head.
It does sound interesting, I normally use the RSI for divergence rather than overbought and oversold so it didn’t occur to me.
I tested the %R on the monthly on all the majors. The difference when you do monthly charts to weekly or daily is that on the monthly it seems to be a complete reversal.
The test was I recorded the price once I got the first signal, then compared the price at the following signal to see if price followed the signal.
Example: I get a buy signal in March 2012 for AUDUSD, price was 0.87. The next signal was in November 2012, another price. I check the price of November 2012, if it was higher than 0.87 then the March signal was accurate.
The limitation of this method is that sometimes price does follow the signal (if I get a buy signal, and price does go up considerably), but by the time I reach the second signal price has fallen below the initial signal. This is unreliable because if I was trading my system I’d definitely have closed the trade for a profit.
But that was the test and I applied for all major pairs. The result was [B]once the ratio reached an extreme level on the monthly chart, there is a 54.7% chance price will see a reversal.[/B]
Safe havens responded worse: USDCHF was 34.7% and USDJPY 50%. USDCAD was successful 36% as well. On the other hand GBPUSD responded 77% of the time. The remaining majors had a success rate of between 60-66%.
So give your test a run, let me know what how you will evaluate the success of a signal and may be comparing the results will tell us what to do next.
