How much can BTC move per Hour? per Day? Weekends? And on Economic events? How about assassination attempt on Trump?
There are different ways to visualize how quickly Bitcoin price can move. You can look at …
- Daily candles
- Hourly candles
- Economic events and Macro
- What happened after assassination attempt on Pro-crypto Trump?
- When trading is thin on the weekends
Why do you want to know this? It gives you an idea how your TP and SL targets in relation to volatility of the asset you’re trading.
‘close_delta_percent’ is base on percentage changes on close prices. So, if your purpose is to determine appropriate stop loss thresholds, bear in mind ‘close_delta_percent’ does not account for the spikes: highs and lows
Daily
On daily time frame, it appears +/- 3% is where most of the action is.
Are weekend’s generally more quiet? Yes indeed.
There are more ways you can look at this. For example if you’re trading the breakout on 30 Aug 2024? The shooting start spiked 4%, still, turned out it was a fake out.
Hourly
On hourly time frame, it appears +/- 50bps capture most of the action.
I can imagine your next question is, how often do you get consecutive outliers like 4th of July pump?
Markus Thielen from 10x research was one of the very few analyst who predicted this.
Economic calendar
Things tend to move more quickly on economic data prints.
Here’s data collected over Mar-July 2023. “pos_neg” refers to “Positive Negative”. For example unemploye_rate, the higher the unemployment the better (i.e. Positive) as we have better odds FEDs will cut rate, and thus supportive of Bitcoin price. Vice versa. It’s a simplistic, naive event classification logic, which is simply WRONG… I haven’t decided how to fix it, as this is a big subject. tbh actually I suck trading events. When I said “Simplistic”, I meant if you try to make money simply “Oh CPI coming down lets long”, it’s really not that simple.
So, here’s how you can look at it. First, you want to see how much price change, hours leading up to the announcement of the print. Then 1st min after announcement, 15min, 1hr, 12hr, 1d thereafter. To trade event, not only that you need to be able to predict if the print will be favorable or otherwise. It also depends on a few other things.
- Most events has a range if you look at historical prints. For example, core CPI came down from 6.5 to 3.3% at the time of writing. The event’s range is 1% to 6.5%. Where current CPI sits within this range? But how will market react to the print? I suspect you can see more positive correlation between Bitcoin price and the print during the walk from 6.5% (top) to 4% (mid range). If you expects a pump after a favorable CPI prints? This will be increasingly invalid as we approach the bottom of range. Example, CPI prints on 11 Jul 2024 was good. Actual = 3.3% vs Estimate = 3.4% and Previous = 3.4%. However, after the print, market pushed to a hi at 59.6, dumped all the way back to 56.5. If you look at 31 May 2024 PCE print, Actual 0.002 vs Estimate 0.003, so the print was good, but at 12th hr and 1d after we’re down about 120bps.
- Again take CPI prints on 11 Jul 2024 as example, inflation coming down good. But as it pushes to 59.6 (Top to bottom a 200bps Doji), we’re approaching major psychological barrier 60k so it became more binary if we could push through it. It didn’t and revert all the way back to 56.5.
Thus to trade events, in additional to making right bet on whether Actual vs Estimates, you need to know:
a) where the print sits within the historical range
b) Is the print breaching important levels, or mid range between important levels? For example CPI 2.9 vs 3.1?
c) if current price nearing any important support/resistance level. You also need to decide if you intend to TP within the first 15 minutes? the first hour? This relates to TP and SL targets. Assuming you made right bet to the event whether it’s bullish vs bearish, it’s still much harder to predict what will happen after 12 hours.
d) Multiple conflicting prints on same day?
e) Interpretation of event: Is higher unemployment supportive of risk assets? Giving more grounds for Feds to cut rate? Wait, what about “Cut for the wrong reasons” (i.e. Recession).
So, wth, is rate cuts now a bearish event? You need look at FREDs historical data points, since WW2:
There are ~a dozen data points if you glance at above, so, that’s not a lot of data points. But generally FEDs cut because they sense economy is already in trouble: Unemployment rising and GDP falling. So one school of thoughts argues: Cuts is thus a bearish event, as recession typically follows. Risk assets generally don’t perform well in recessions. The counter arguments (VirtualBacon) are:
- If you look far enough, recessions always comes.
- Who declares “Recessions”? It’s National Bureau of Economic Research (NBER). They are typically very late.
- Outliers: If you look at dot-com, GFC and COVID the most recent events (This is already 30 years), things are already broken when FEDs cut rate. From view point of 29 Sept 2024, ask the question: Are things broken today? Are we in similar situations as previous three crisis's? The nature of the beast may have already changed fundamentally. In olden days, Warren Buffet invested based on firms’ fundamental. Today, we trace liquidity cycles. So how relevant are events before 1980s, when monetary policies were much more grounded?
Even if you can accurately predict the print, predicting how the market actually respond is not straight forward. You need keep watch as economic prints (GDP and Unemployment) unfold, and make adjustments.
f) Other things happening? Have you looked at the flows?
Please dig into the data file shared. You can see first minute, Bitcoin prices can move 150–200bps in 1st minute, 15min, 1st hour after print announcement. What happens next day is harder to predict. Example, 3 May 2024 unemployment print was followed by 650bps pump, and that’s after the sell offs from 30 Apr 2024 to 1 May 2024
After everything I babbled, I admit I don’t know how to trade events, as first you need predict Actual vs Estimates. How do you go about doing that? Model it? And, are we complicating things? Do we need to actually predict events outcomes? Why not simply gather aggregated orderbooks, and trade orderbook imbalance? Trade what you can see.
However, above information is still useful: If you have an open trade, and event move against you. How much can you expect it’d move against you? Again, it relates to your SL targets.
What happened after assassination attempt on Pro-crypto Trump?
On fateful day 14 July 2024, that’s a 120bps pump, within first first five hours.
Then give it two more hours, the bear market was canceled.
When trading is thin on the weekends — 0xAnn has something to say!
Now things get more interesting, from an article from 0xAnn
Any truth in this? Lets check daily candles. In my code, look for “close_vs_2_days_ago_gmt_bps”.
x-axis of histogram in bps.
So in conclusion, 0xAnn is correct price move up generally over weekend when market in general uptrend, of 2022,2023,2024, seems only Jan-Jun 2024 with a hitratio of 60%. This will not work in bear market.
You can probably improve your odds by employing additional signal. For example use 30d EMA as north star. Above you trade. Below it you don’t trade.
You can easily estimate return by:
daily_candles['pnl'] = 100000 * (1/10000) * daily_candles[(daily_candles.dayofweek_gmt==6) & (daily_candles.year==2022) & (daily_candles.close_2_days_ago>daily_candles.ema_close_2_days_ago) ]['close_vs_2_days_ago_gmt_bps']
daily_candles['pnl'].sum()Above assume you put in USD 100k. And return pnl is on 1x leverage:
- 2022: +4%
- 2023: +15%
- Jan — Jun 2024: +12.8%
Above is just rough estimate. It didn’t discuss things like hard SL (Stop Loss) and price spikes handling. I didn’t calculate Sharpe, but given hit ratio’s, probably not my favorite trade.
The Code
It’s all in Google Colab.
https://colab.research.google.com/drive/1OU_2Sl-6jpQPrNyw9EFrhgIpDe4M0A6m#scrollTo=6pZCETGwBcU5