Dimsky's Journal: 100 x 100

This is gonna be a good thread!

image

1 Like

Added a few fields, revisited EURUSD observations to populate data and did USDJPY. Failed to mention previously that 100 pip observations are on a minimum 2 candle range. 100 pip + daily spikes are ignored.

Also obvious oversight on the trading TF. With this account size I cannot trade above H1 where my default is to open with an SL of either 15 pips or 1xATR (higher of 2) and then trail 1.5-2xATR after moving to BE. The D1 observations are still relevant since I’ll be switching to a triple screen of D1 / H4 / H1 now instead.

It’s time consuming but needed. I’ve never looked at charts as closely as I do now and this approach helps. It’s suits the way I look at things. Might discard it over time with greater experience when things start to feel more intuitive but I doubt it.

Also wasn’t entirely happy that I was only dedicating so much time against D1 only. But then I realized I’ll be going over H4 & H1 with a fine tooth comb when I back test it against the predetermined criteria from this exercise.

PS - Happy New Year to anyone catching this! :partying_face:

211218_Significant trends_Major CP.xlsx (31.7 KB)

1 Like

Caught it…and wishing you a very Happy and Rewarding New Year 2022! :+1:

2 Likes

Wish you the same!! :partying_face:

1 Like

Happy New Year @darthdimsky.
Good information & journaling.
But who uses indicators now than only PA? Bars/Candles are footprints of trades.
Money is somewhere else, that people will not tell you & difficult to catch tho. Happy Analysis & Learning.

1 Like

@petrogold - Thanks for the well wishes. While I’ve gotten used to looking at charts the past year I’ve always felt comfortable looking at data on spreadsheets. Indicators, to me, quantify PA in numerical terms. I don’t shy away from using PA. I just don’t rely on it solely. Tend to use it as a further confirmation for entries/exits.

Journal
Decided to take a break from staring at the D1 charts and attempted to code something in MQL5 (incomplete). Also thought I’d look at the H1 data for all the US pairs to get a better idea about the hourly activity because I’ve never really paid heed to the TYO session. Only spent time on the LON and NYC sessions. Wanted to know if I was being an idiot.

Charts show an average change in pip values and tick volumes for Calendar Year 2021 by hour. Times are in GMT+2 (MQL5), which translates to the following session times:

  • TYO = 0100 - 1200
  • LON = 0900 - 1800
  • NYC = 1400 - 2200

Notes:

1. Tick volume is according to liquidity provider against my broker. May not be an accurate representation of tick data for all LPs out there.
2. DST isn’t taken into account. I’ve never lived with DST and calculating it confuses me & gives me headaches. So there might be a +1hr skew for a portion of the data.
3. Certain months traditionally have had more activity than others so there could be greater deviations in certain months.

Overall tick volume gives a more accurate picture than hourly change to tick vol. It also shows a strong correlation b/w tick volume and hourly change in pips. Following charts I found are less confusing.

Man, sad to hear about Betty White passing away. RIP :cry:

GBPUSD done. AUDUSD next. Without quantifying anything concretely and with just looking at D1 charts I’m able to determine:

  1. The SMA is a great entry indicator. This isn’t surprising with a lot of forum articles and TA/PA books giving a preference for EMAs (EMA20 in particular). Am finding that lower D1 volatility around (50-60 pips) tend to conform to SMA10 while higher volatility (80-100 pips) tend to conform to SMA20.
  2. A Bollinger band exit on the Std Dev 1 can be a reliable exit indicator. With pairs of lower volatility just following this blindly will lead to premature exits because bands have a higher constriction and you risk being whipsawed out of a trend. Perhaps a first exit after a relevant countertrend candle?
  3. Preliminary looks on the ADX on the W1/D1 TF are inconclusive. Perhaps there will be more consistency in the lower TFs. Also wondering if I’m looking at the ADX all wrong. I’ll know soon.

Had also noticed spread data in the initial data export. Determined it was largely unreliable since 69% of the data for the 7 USD pairs consisted of a zero spread.

My broker has a zero spread accounts and perhaps the data is accurate for those accounts. Not relevant for my Standard account though, which is also where the export data is from.

Also used the data to determine whether TYO or LON sessions could be used as a leading indicator for the the movement of price for the following sessions. It was inconclusive. Still suspect there could be a correlation b/w LON & NYC sessions, which I’ll look at later.

211218_Significant trends_Major CP.xlsx (42.5 KB)

Was re-reading portions of Ashraf Laidi’s book because I couldn’t apply anything in there during the time my laptop was out of commission.

While going through the first chapter on Gold I remembered watching a real vision finance interview with Tony Greer that discussed BTC and it’s possible risk off traits. I wanted to know if there were any numbers that justified it. The only place I thought of looking were maybe COT reports. I don’t know futures and didn’t know if there was a BTC future. Luckily found this that indicated it’s availability.

Downloaded a few of the variations of the reports and figured out which one the Insider Week website was sourcing it’s data from (“Futures Only” here), how the numbers were derived and why they were derived that way.

It’s a steep learning curve for someone with no knowledge of futures. But satisfying to understand, ateast the parts I wanted to know. Some data for future reference if I have to trawl through the data again:

Decided to look at the correlation for the entire 2 year period and it’s inconclusive. Just a cursory glance indicates no relationship/weak relationships between Gold and other instruments

Looking at 2 years worth of data doesn’t tell me whether there were changes to the the relationship mid way. Meaning there’s a possibility that there were periods of strong correlationship between XAU and the other instruments.

Implemented a rolling 12-week correlation to to gold. No real significance to the 12 week duration. The books outlines a 6 month rolling correlation but for a 7-8 year span. It showed times when BTC and Gold were strongly correlated, FEB-APR & again in JUN-JUL

The rest of the correlations to the AUD & JPY didn’t make a lot of sense. Especially the positive & negative correlations, respectively, in the first half of the year. That made no sense.

The exercise was meant to determine whether BTC is a risk off assett but I was only considering Gold as the only risk off indicator so far. That wouldn’t make sense if some of the capital/contracts that would traditionally have gone to Gold are now taken up by BTC instead. This is evident if you standardise the open interest for XAU & BTC (07JAN20 = 100) and plot them side by side.

What the chart clearly shows is that BTC contracts, at the end of 2021, are at ~231%, compared to the open interest figure on 07JAN20, while XAU contracts in comparison are ~65% of the corresponding volume

For a more meaningful correlation it’d make more sense to look at the aggregated open interest for BTC & XAU, which if considered for the entire 2 year period would look like:

When looking at overall numbers the near zero correlation with the strongest risk off currency and a positive correlation of 0.92 with a currency that has the most hawkish central bank makes no sense. The rolling 12 week exercise was applied again after applying correlations to the aggregated XAU/JPY open interest with more encouraging behavior.

Number of takeaways from the graph:

  • The NZD & AUD show a strong correlation to each other, atleast in the first 3Q of the year. That’s consistency is encouraging.
  • Why do the risk on currencies have a positive correlation to risk off assets (not just the aggregated XAU/BTC but also the JPY)? Were the AUD/NZD that weak in the periods MAR - early MAY and late JUN - SEP?
  • The risk on currencies show the same dip to no relation that Gold showed to BTC for the same time period, between MAY & JUN
  • While the NZD is more inconclusive toward the end of the year, its encouraging to see the strong positive correlation against the JPY and the strong negative correlation against the AUD at year end, which gives credence to the theory of BTC of being a strong risk off asset.

Some of the anomalies can be explained as folls:

  • The dip to no correlation of XAU to BTC and later XAU+BTC to NZD & AUD - The timeline is significant. This is when all hell broke loose with BTC after the ransomware attack on the colonial pipeline on 07MAY. That began a cascade of events, like the US government’s crackdown on BTC, Musk’s announcement to stop accepting BTC payments for TSLA & China’s eventual ban on cryptocurrency mining. Further confirmed by the sudden dip in daily hashrates for BTC in MAY.

source - Bitcoin Hashrate Chart

  • The NZD’s non correlative behavior in Q4 - Does this coincide with the drastic coronavirus lockdowns that began with the discovery of it’s first cononavirus case since FEB? I remember that’s when we were expecting the RBNZ to increase it’s interest rates. The first Central Bank at the time. Though I’m keen to pin the lack of correlation to the unexpected market environment caused by the drastic measures taken for the coronavirus lockdown I’m a bit hesitant to say that’s the certain cause without more data points for validation.

  • The positive correlation between AUD/NZD (Risk on) and XAU+JPY (Risk off) Open Interest for MAR-MAY & again from JUL-SEP is a huge cause for concern. That’s just not supposed to happen, atleast not to the best of my limited knowledge. Recap of that behavior on the same chart with remarks:

Decided the best way was to determine the validity of the behavior was to apply the same correlation to the indices for the same currency pairs. In the absence of an instruments like the DXY equivalent (atleast not that I’m aware of atm) for the USD. I decided to use the TWI or the EER equivalents. Thankfully there’s a centralized repository for the EERs at the BIS here. The only issue is that this might be real when I’d thought till date that nominal effective exchange rates was the way to go.

Is there a positive correlation between the JPY and AUD/NZD in the effective exchange rates?

Unlike other correlations I decided anything > -0.4 is significant. Simply because we expect risk on/risk off instruments to mirror. This period overlaps with the anomalous zones determined by the opened interest positions in the COT data, which, for now atleast, doesn’t invalidate the approach I took. But to even see a correlation > 0.4 between the JPY and AUD/NZD at some points is really surprising and cool. Didn’t expect that.

Conclusions/Takeaways:

  1. At present BTC & Gold is used as part of portfolios for Risk off behavior and the COT data can quantify that.
  2. Applying a rolling correlation can give a lot of insight into an instrument’s behavior in light of changing market conditions. Credit for this idea in this instance is with Mr Laidi in his book where he demonstrated it:
    Ashraf Laidi snippet
  3. Risk on instruments can at times have positive correlations with risk off instruments! :face_with_raised_eyebrow:
1 Like

On another note - Came across this fascinating study done by the CFTC about BTC trader profiles here:

The most interesting bit was the fat tailed distributions, where a majority of traders had BTC as <20% or >80% of their total portfolio. Some very interesting observations there.

CFTC BTC participants

100 pip+ D1 trend analysis

AUDUSD done. This was a very tiring exercise. I hated it. I hate manually plotting and gathering data points. I procrastinated over 2-3 days because I hated it so much. Pain in the frickin @$$. This is why I learnt MT5 for that csv in JUN. This exercise has only motivated me to go back to it and complete it. I will take a break from the H4/H1 analysis and compile that code again.

211218_Significant trends_Major CP.xlsx (50.1 KB)

D1/H4 insights

Downloaded H4 & D1 broker data and decided to run some numbers with the H1 data I had earlier. I wanted to know how the lower TF moves with the overall move for the day. For e.g. if the D1 was bearish how many of the lower TF candles proved to bearish on average over the year. One of the reasons was to determine if prior sessions could be used as leading indicators for PA.

Another reason was the boredom from the 100 pip trend analysis. I used the prospect of crunching these numbers as a carrot stick to get myself to finish that exercise quickly.


Takeaways

  • No surprise that that the LON and NYC sessions generally more or less determine how the CPs on average move for the session.
  • What would make it more relevant is if a combination of H4 candles in the same direction generally coincide the overall direction for the D1 candle. For e.g. it’s found that when 60% of the 0000 & 0800 candles are bearish when the D1 candle is also bearish. This is possible. Might look at it later.

Overall doesn’t appear to be a huge surprise. Each of the respective currencies are the busiest at their respective sessions. Perhaps it could do with an average tick volume overlay to give more meaning?

D1/H1 insights

Identical exercise with D1/H1 instead.

Takeaways

  • EURUSD: ~58.5% of 0000hr candles tend to go in the direction of the D1 candle. In raw % terms that could be statistically significant.
  • USDCAD: ~57% during the TYO session @ 0200. Also remarkably is an inverse relationship with the 0100 candle that indicates that 55% of the time this candle indicates the opposite direction of the D1 candle.
  • USDCHF: ~58% & 55% of 0000 & 0100 movements respectively coincide with overall D1 movement. This could be a significant indicator of risk-on/risk-off behavior by the big JP players.

The USDCHF chart raised more questions. Would adding tick volume give more context? Previous numbers already show a strong correlation between pip movement and tick volume.


Avg tick volume registered is ~24% more (at ~714) for 0000hrs when H1/D1 candle directions match. Great stuff. Bit tired atm to do the rest of the charts but EURUSD was also really interesting


Tick volumes @ 0900/1000/1100 register an increase of ~500(15.4%)/~350(8%)/~400(10.6%) respectively. Possibly a significant leading indicator heading onto the NYC session. Exciting stuff!

Fundamental data

FXStreet’s economic calendar has an option to download as csv. I was super excited to find this till I looked at the raw data.

No figures. none. zilch. So disappointing. Atleast it enables some insight:

FXStreet_AUD High

Felt a strong burning desire to dump everything and learn Python for online data scraping when I saw this initially. Thankfully I had already found out it might be possible on MQL5 back in JUN, so that’s another incentive to go back to coding in MQL5. I will do my best to avoid manual data collection for key fundamental data.

Psychology

Huge gaping issue atm. I have pored over technical books, read up on statistics, economics & fundamentals. But next to nothing on psychology. This is despite knowing how weak my mental disposition was when I was playing poker. Though most of the lessons I learnt then, I feel, are ingrained in my DNA now, I also know it’s still lacking. It’s a muscle I stopped working on a long time ago. That was a mistake.

Going to devote a majority of my time to developing the MQL script, followed by psychology. Will work in parallel with the H1/H4 analysis as a 2nd priority.

1 Like

Hi @darthdimsky

Thank you for posting this heavy work. It must be the nerd in me, but I have a huge admiration for people like you who do original research. On the one hand you may be forgiven for thinking that there are loads of rocket scientists at JP Morgan who get paid $150K per year to do this sort of stuff.

On the other hand, it has been my experience in industry that half the people charged with using automation programmes to extract intelligence from raw data just don’t bother to look at context or check their work with some manual calculations and just downright reality of the universe. I had a really difficult time last year to try to find the way to show the head of BI in a very large company that his data extracts from ServiceNow (helpdesk configuration management database) were shoot - garbage in, garbage out. It would have been far easier just to say that he should have understood his own corporate data after 20 years instead of looking at pretty patterns for the CxO consumption. Together we agreed to inform the powers that be that we had “misinterpreted the data extract criteria” and used incorrect data. That saved a lot of faces.

I am looking forward to seeing some real examples of both API and RPC calls as I study blockchain for the next few weeks. I want to look under the hood of modern decentralized networks and their ability or otherwise to cope with flash demand - like response to a new offering, or just to contain spam on low gas networks. I suspect, somehow, that at a lower level we have a subject matter of common interest.

1 Like

I wish I could claim originality but I took a lot of inspiration from reading Kathy Lien’s book on this one. Here’s an example of her analysis for the Asian session

Kathy Lien_Asian session

I think there’s an element of truth though to large firm and pro traders looking at price action in a similar manner. Especially in this day and age where data science is as popular and tools accessible. The above bar chart I came across in the 2nd edition of the book, that was published in 2008. Here’s one way I want to analyze economic news releases:

Excerpt from “Day Trading & Swing Trading the Currency Market”, 3rd edition
Kathy Lien_Economic calendar impact Kathy Lien_Economic calendar impact2

I’ve mentioned Kathy Lien so many times in so many places in BP that I’m scared of being reported for promoting her. I found a very similar line of thinking with Linda Raschke and she was a pit trader in the 80s who transitioned to fund manager. Edit: Kathy Lien used to work for JP Morgan, followed by FXCM.

The disappointing thing for me is I haven’t come across any analysis dissecting price action like this. Maybe it’s there and I’ve not stumbled on it yet. It’s disappointing because I want to bounce ideas and improve my knowledge in this space but am unable to. I’m constantly worried that I’m looking at it the wrong way. Edit 2: Actually that’s exaggerated. It’s not worrying if you can validate those findings using another data set, like I did with the COT data findings with the broker data. More like I’m worried if I’m missing out on another way of looking at the data I already have in my possession. That’d be tragic.

I know next to nothing about cryptos because I’ve consciously avoided it. Most I feel are trading it because of the hype around it. But I’d be very interested to read any findings in your analysis, simply because it’d be data driven. I haven’t come across anything like it in BP so far.

Hi, thanks for the response. You are too humble. I still think you are doing some very useful fundamental statistical analysis. It was a long time ago now, but I attended a weekend forex course in London hosted by Jimmy Young, who had been trading majors for over 20 years after he left his corporate career at various banks on their Forex desk. Jimmy made a point, just like you are doing, about his analysis of past 20 years of raw data, and came to the same conclusion as you - there was more than a 60% correlation between whatever it was he had measured, and so he always used last trading day closing price as his baseline. Within the first 10 minutes of London open, he chose to trade with trend (I can’t remember if Monday’s trend needed to be same as Fridays, or what was the indicator he used to open a trade, but he had a rule for entry, stop loss and take profit. It was based on a 60%+ probability of being able to catch at least 70% of x pips, x being the statistical average of the 60% probability he caught the trend in the right direction. If I recall rightly, by going for 75% of the pips in the right direction, his win / loss ratio was about 65%. We did not get any notes to take home and his presentation pack files emailed after the presentation did not cover the detail he lectured about.

The link below is to the same guy who hosted the London conference. Attendance was a (reasonable) $300.

Edited. Oh, about the cryptos, and for trading them, I don’t think there is anything magical about cryptos that doesn’t occur with the majors and crosses. It’s just that there are 10,000 of them, and most of them pair with either BTC or one of the USD stable coins.

I do follow a lot of fundamental analysis about cryptos on Youtube, always make sure I have read the whitepapers of those I choose to trade, and the reason I really like trading them is because they are so volatile, I do not use leverage and hence do not use a stop loss. If and when they collapse to zero, it is never overnight. It is always a long tail down to a satoshi or two.

1 Like

What I love about the whole Jimmy Young bit (and I hope I don’t sound patronizing) is that he gives #s that you can use to determine whether his argument holds water or not. He’s basically giving you the tools to debunk his own method.

I’d be able to download broker data and apply a similar methodology too after hearing him out. And that’d give me the opportunity to find out if it works for the current market conditions. Perhaps it works for some instruments and not others? That too can be tested. That’s why I like it.

I really meant about being ignorant about cryptos. I knew whitepapers existed but I didn’t know there was fundamental analysis around it. First time hearing about it. That’s pretty cool. Do you have a go to YT channel you found reliable?

Hi,

By far the most reliable source of information is linked below. It’s Guy from Coin Bureau. I thought I had understood what I needed back in June 2020 when I created my first investment / trading plan based on crypto “types”. That was just scratching the surface. I think by now, though I have not measured it, I have probably spent more than 1,000 hours doing some R&D on all things crypto.

I don’t wish to tempt fate, but last night I traded a NFT for a 15X. I did not think that possible, and didn’t really want to sell it, but as you probably agree “everything has a price”, so I put it up for auction on a NFT site at about 10 times its floor price - and it sold due to its rarity. I still can’t get my head around what kind of nutter pays a 12X (it was a 15X in USD terms) when the buy price is in plain sight in the contract, and I’ve owned it for just less than a month.

To date I have bought and sold over 10 NFTs, and all the others were within a small range. But this one last night really made me think - is this the year I give up the day job? :rofl:

I have literally hundreds of links in my OneNote folder for Crypto. Below is a snippet of the folder contents of some of the cryptos I have researched since June 2020.

I just finished a re-definition of my allocation last night, and it now contains up to 30% of value to be diverted from longer term holding to shorter term Liquidity Pair (LP) and NFT trading. Three months ago, both LP and NFT trading were not even on the radar. Today, they constitute almost 50% of our portfolio. Too important to ignore. I have @ponponwei or @ria_rose to thank for that. One of them gave me push to think about playing “play to earn” games, which I dismissed outright, and then decided to look into after talking to an ex-flat mate of my son who got a job inside the GameFi industry. The rest is history.

I am very fortunate to have the time to go and do research into what others think are very off the wall ideas. I now have the basis for a new business, because I think I can scale the identification and filtering of NFT trading without staff needing to understand crypto or even open a crypto account.

The reason I am so clear about spending time pursuing this is because I learned some of these skills playing Runescape with my sons, and during the past three years, I have been practising price discovery methods in that game to reach a “gold pieces” balance of 900 million GP. I haven’t played for a while, but all that fun time has become a serious part of my trading plan for NFTs. Strange how life works out. :crazy_face:

1 Like

I think I remember Coin Bureau when I initially read up on Cryptos in 2017. I think it was the most reliable source of crypto information at the time. I didn’t know about it’s YT channel. Will give it a look.

Yea, Runescape’s free market has a lot of interesting dynamics. The price a 3rd age bow has skyrocketed to the 2B range atm I heard recently. This was 3-400m in 2019 atleast. There are some really good flippers in game that also trade in real life markets I think. Sadly I spent time levelling up non combat skills and not enough merchanting so I don’t have any in-game skills that help in trading.

Then you would like DeFi Kingdoms. The NFTs in that game are Heroes (like Pokemon trading cards) who you send on quests.

LOL I think I would which is why I’m going to avoid it for now. I have a very difficult time switching off something I take an interest in. That trait has been helpful for trading but with games it’s proven to be a time sink. I was skilling as an ironman (manufactured/harvested most skilling requirements) on a normal account. My OSRS stats:

2022-01-19_00-46-10

I want to make trading work full-time. There’s a lot to do and gaming, in my case, will get in the way. If and when I succeed (swing trading on larger capital) I’ll go back to OSRS I think. I feel disappointed I couldn’t finish some in-game objectives.

Journal:

Penning some thoughts I’ve been processing on the downtimes.

  • Need to implement a grading system in my journal. Possible solution here and another possible one for ref in “The New Trading for a Living”.
  • Get back on the daily notion grind after 100 pip analysis
  • Resume and keep the Anki grind. No excuses.

Current to do:

  • MQL5 script
  • Read Trading in the Zone & Mental Game of Poker to start drafting a mental routines/exercises
  • 100 pip exercise continuation on H1/H4
  • Incorporate a journal grading system
  • Brainstorm a method to quantify mental state, at trade, within journal.
1 Like