PyPriceAgent keeps track of a history of stock and crypto prices and sends you alerts when various Technical Analysis type events happen on the charts.
It keeps a csv and html-table file with latest prices up to date, for import into a spreadsheet.
It also allows some back-testing.
It'll keep track of bets you place and let you know when they win/lose/expire in the email each day too.
(C) pre@dalliance.net, but I don't care, do what thou wilt.
For stock prices, you'll need an API key from worldtradingdata.com They have daily rate limits for free accounts. The code here tries to minimize calls by caching vigorously, but if you add too many stocks you'll hit their limits.
It'll need these modules.
sudo pip install tabulate #Display results in tables
sudo pip install pandas numpy #moving averages, maths on series.
sudo apt-get install python-matplotlib #plotting graphs
You'll want to place a file "secrets.py" in the same directly as pyPriceAgent.py, to hold API keys and SMTP details etc. There's an example one with dummy data at secrets_example.py
You should set a cron to run it some time after market close: Something like this probably:
00 21 * * * python -u /home/user/py/pyPriceAgent/pyPriceAgent.py --email true -l2 >> /home/user/log/stocks.log 2>&1
The "tickers.txt" file lists the stocks, one per line, in a format like you'll get exporting a trading view list. EG: NYSE:IBM
The "bets.json" file is created to track any reminder bets you place.
The "caches" directory holds API request-results etc.
The "history" directory holds our daily price-history for each stock
Both dirs must be writeable, all the data is just JSON dumped into files.
You may adjust the default four MA/EMA values in the tickers.txt file, with 4 or 8 numbers separated by spaces after the ticker name. First the EMAs short to long, then the MAs short to long.
EG:
nasdaq:amzn 25 60 95 180 25 60 95 180
By default we write files to ".new" and then rename them to overwrite the old version. I'm told this may not work on Windows. You can override the behaviour with secrets.py but if you do that don't CTRL-C as it runs or you'll keep destroying your data.
Run --help to see the CLI params,
You can add a new bet at the CLI:
./pyPriceAgent.py -t LSE:TUI -B Price/target/stop/days/confidence/startDate
Defaults exist for them all, type "X" to get:
Price -> Previous Day's Close
Target -> +15% on price
Stop -> -5% of price (or +5% if target<price)
Days -> 20
Confidence -> 33
StartDate->Today
You'll have to just edit the JSON in bets.json if you screw up. There's more planned here that I haven't gotten to yet.
Confidence is more or less ignored, but I'm sticking to <50 for bets I don't take and >50 for bets I do actually put money on for now. I'd suggest you do similar.
There is copious comments in the source-code for more about how it works and what the various options are. Skimming the comments may well answer many questions.
Just fetch today's prices and print the email alert:
./pyPriceAgent.py
The output shows the email it'd send. In this case no bullish alerts, but two worrisome RSI daily indicators on Bitcoin.
From: example@gmail.com
To: example@your.email.moo
Subject: Daily Stock Summary
2020-02-10
Nice looking things this time:
And some things looking bad:
-1: BTCEUR.COINBASE RSI Daily falling
-1: BTCUSD.BINANCE RSI Daily falling
Course, if printing the email content isn't good enough and you want to actually SEND it, then hopefully your SMTP details are in secrets.py then:
./pyPriceAgent.py --email true
And it'll send it. The output to the console will just look the same though, but an email should arrive in your inbox too.
Your ticker-list will likely grow bigger than this example, and sometimes you might only care about one ticker. Especially when back-testing later.
We can filter to show only some or one ticker.
./pyPriceAgent.py -t BTCUSD
Limited to only BTCUSD we only see things about BTCUSD
From: example@gmail.com
To: example@your.email.moo
Subject: Daily Stock Summary
2020-02-10
Nice looking things this time:
And some things looking bad:
-1: BTCUSD.BINANCE RSI Daily falling
Or if we wanted to see everything about BTC we could just include that, or just "LSE" to see only London exchange stocks etc.
What about in the last week? Use -b or --backtest
./pyPriceAgent.py -b 7
Now it tells us about a few more things, and tells us how many days ago they happened. There was a Red-9 on DBCBTC yesterday, and four days ago the EMA cross meant that the moving averages were sorted into fully-bullish order.
From: example@gmail.com
To: example@your.email.moo
Subject: Daily Stock Summary
2020-02-10
Nice looking things this time:
1: DBCBTC.HUOBI 1 days ago: SEQ Red 9 d
1: BTCUSD.BINANCE 4 days ago: EMA Sort Went Bullish
1: IBM.N 3 days ago: EMA Sort Went Bullish
And some things looking bad:
-1: BTCEUR.COINBASE RSI Daily falling
-1: BTCUSD.BINANCE RSI Daily falling
-1: IBM.N 1 days ago: RSI Daily falling
If we back-test even further then we start to get to triggers that happened long enough ago that we know what the price did after that trigger. So we can start to build up some statistics
./pyPriceAgent.py -b 1000
The output now contains tables showing the average gain in price of the assets after a TA event happened. In this case in the third line of the table we can see that after a Bullish Sequential event, the price was up an average of 0.3% 1 day later, 0.4% two days later, 1.2% 10 days later etc.
Below that is a table telling us what proportion of the triggers were followed by a gain of 10% in the next 1 day / 5 days / 10 days.
We can see that a bullish RSI signal was followed by a 10% rise within ten days 3 out of 40 times, and within 25 days it'd hit that 10% target ten out of 40 times.
Note that the email summary output only contains one row for each ticker in the list. It shows the one in which the most triggers happened on any given day. TSCO had both a sequential red 9 daily and an increasing RSI daily signal 358 days ago.
From: example@gmail.com
To: example@your.email.moo
Subject: Daily Stock Summary
2020-02-10
Nice looking things this time:
2: TSCO.L 358 days ago: RSI Daily up, SEQ Red 9 d
2: AMZN.O 128 days ago: RSI Daily up, SEQ Red 9 d
2: DBCBTC.HUOBI 46 days ago: SEQ Red 9 d, SEQ Red 9 w
2: IBM.N 321 days ago: SEQ Red 9 d, RSI Weekly up
1: BTCUSD.BINANCE 4 days ago: EMA Sort Went Bullish
1: DGBBTC.BITFINEX 40 days ago: SEQ Red 9 d
And some things looking bad:
-1: BTCEUR.COINBASE RSI Daily falling
-2: BTCUSD.BINANCE 239 days ago: RSI Weekly falling, SEQ Green 9 w
-2: TSCO.L 201 days ago: SEQ Green 9 d, SEQ Green 9 w
-2: AMZN.O 204 days ago: RSI Daily falling, SEQ Green 9 w
-2: IBM.N 240 days ago: RSI Daily falling, SEQ Green 9 w
Average Gain After Signal:
Signal Direction 1 Bar 5 Bar 10 Bar 15 Bar 20 Bar 25 Bar
-------- ----------- ----------- ----------- ----------- ------------ ------------ ------------
seq_w Bull -0.3 % / 49 -2.3 % / 49 -4.0 % / 49 -3.6 % / 49 -3.7 % / 49 -4.2 % / 49
rsi Bull +0.0 % / 40 -0.8 % / 40 -0.9 % / 40 -1.4 % / 40 +0.8 % / 40 +1.2 % / 40
seq Bull +0.3 % / 60 +0.4 % / 60 +1.2 % / 60 +0.4 % / 59 +0.4 % / 57 +0.2 % / 57
multi Bull -1.1 % / 13 -2.6 % / 13 -2.2 % / 13 -3.2 % / 13 -3.4 % / 13 -3.5 % / 13
emax14 Bull -0.7 % / 20 -0.5 % / 20 +0.8 % / 20 +0.7 % / 19 +0.8 % / 19 +1.3 % / 18
rsi_w Bull -0.7 % / 40 -1.0 % / 40 +0.9 % / 40 -1.0 % / 40 -1.9 % / 40 -2.2 % / 40
emasort Bull +1.0 % / 22 +1.4 % / 22 +0.3 % / 22 +1.7 % / 21 +3.2 % / 21 +4.0 % / 20
seq_w Bear +0.4 % / 99 +2.1 % / 99 +5.4 % / 99 +7.1 % / 99 +7.1 % / 99 +8.9 % / 99
rsi Bear +0.5 % / 80 +0.5 % / 80 +1.9 % / 80 +2.2 % / 77 +3.3 % / 77 +4.5 % / 76
seq Bear -0.1 % / 66 +0.7 % / 66 +1.2 % / 66 +1.6 % / 66 +2.3 % / 66 +1.7 % / 65
multi Bear +0.9 % / 26 +3.1 % / 26 +9.4 % / 26 +12.7 % / 26 +10.6 % / 26 +12.5 % / 26
emax14 Bear -0.0 % / 17 -1.2 % / 17 -0.6 % / 17 -0.7 % / 17 -0.9 % / 17 -1.5 % / 17
rsi_w Bear +0.6 % / 98 +3.0 % / 98 +3.9 % / 98 +4.9 % / 98 +4.8 % / 98 +3.8 % / 98
emasort Bear +0.2 % / 19 +1.6 % / 19 +1.9 % / 19 +3.3 % / 19 +1.4 % / 19 +0.8 % / 19
Number that Hit +/-10.0% gain/loss in bull/bear by:
Signal Direction 1 Bar 5 Bar 10 Bar 15 Bar 20 Bar 25 Bar
-------- ----------- ------- ------- -------- -------- -------- --------
seq_w Bull 0 / 49 0 / 49 0 / 49 0 / 49 0 / 49 0 / 49
rsi Bull 0 / 40 2 / 40 3 / 40 4 / 40 8 / 40 10 / 40
seq Bull 0 / 60 4 / 60 6 / 60 6 / 59 7 / 57 9 / 57
multi Bull 0 / 13 0 / 13 0 / 13 0 / 13 0 / 13 0 / 13
emax14 Bull 0 / 20 0 / 20 1 / 20 1 / 19 3 / 19 3 / 18
rsi_w Bull 0 / 40 0 / 40 4 / 40 6 / 40 6 / 40 6 / 40
emasort Bull 0 / 22 1 / 22 1 / 22 3 / 21 4 / 21 4 / 20
seq_w Bear 0 / 99 0 / 99 0 / 99 0 / 99 0 / 99 2 / 99
rsi Bear 0 / 80 0 / 80 2 / 80 3 / 77 5 / 77 9 / 76
seq Bear 0 / 66 1 / 66 2 / 66 5 / 66 8 / 66 10 / 65
multi Bear 0 / 26 0 / 26 0 / 26 0 / 26 0 / 26 2 / 26
emax14 Bear 0 / 17 0 / 17 2 / 17 3 / 17 3 / 17 4 / 17
rsi_w Bear 1 / 98 1 / 98 4 / 98 7 / 98 10 / 98 16 / 98
emasort Bear 0 / 19 0 / 19 1 / 19 1 / 19 2 / 19 3 / 19
We can limit which tests we decide to run, or include some of those not run by default.
Imagine we wanted to know how long it takes Tesco's price to go up by 20% after a bullish weekly RSI signal in the last 2000 days.
./pyPriceAgent.py -t TSCO -b 2000 -c rsi_w -p 5
The output tells us that 2/45 times Tesco went up by 5% within 5 days, and 10/45 times it was up that high within 25 days.
From: example@gmail.com
To: example@your.email.moo
Subject: Daily Stock Summary
2020-02-10
Nice looking things this time:
1: TSCO.L 281 days ago: RSI Weekly up
And some things looking bad:
-1: TSCO.L 196 days ago: RSI Weekly falling
Average Gain After Signal:
Signal Direction 1 Bar 5 Bar 10 Bar 15 Bar 20 Bar 25 Bar
-------- ----------- ----------- ----------- ----------- ----------- ----------- -----------
rsi_w Bull -0.8 % / 45 -2.6 % / 45 -2.0 % / 45 -1.1 % / 45 -2.9 % / 45 -5.1 % / 45
rsi_w Bear +0.6 % / 35 +1.9 % / 35 +0.8 % / 35 +2.1 % / 35 +1.8 % / 35 +0.7 % / 35
Number that Hit +/-5.0% gain/loss in bull/bear by:
Signal Direction 1 Bar 5 Bar 10 Bar 15 Bar 20 Bar 25 Bar
-------- ----------- ------- ------- -------- -------- -------- --------
rsi_w Bull 0 / 45 2 / 45 9 / 45 10 / 45 10 / 45 10 / 45
rsi_w Bear 0 / 35 0 / 35 1 / 35 4 / 35 10 / 35 10 / 35
We can examing what happened to the price once more than one check triggered on the same day. We limit our checks to only those we care about, and enable the "multi" check.
If there was a sequential-nine on both the daily and the weekly candles at once, what happened to the price?
./pyPriceAgent.py -t TSCO -b 2000 -c seq,seq_w,multi
Well, with Tesco it's only happened three times in the last 2000 days. The two times it was a bullish signal, it went up an average of 0.5% the next day.
When it was a bearish signal it fell 1.3% the next day, and was down 8.7% after 20 days.
From: example@gmail.com
To: example@your.email.moo
Subject: Daily Stock Summary
2020-02-10
Nice looking things this time:
2: TSCO.L 1067 days ago: SEQ Red 9 d, SEQ Red 9 w
And some things looking bad:
-2: TSCO.L 201 days ago: SEQ Green 9 d, SEQ Green 9 w
Average Gain After Signal:
Signal Direction 1 Bar 5 Bar 10 Bar 15 Bar 20 Bar 25 Bar
-------- ----------- ----------- ----------- ----------- ----------- ----------- -----------
seq_w Bull -0.4 % / 35 -1.6 % / 35 -3.7 % / 35 -6.3 % / 35 -8.1 % / 35 -8.6 % / 35
multi Bull +0.5 % / 2 -0.5 % / 2 -2.5 % / 2 -7.7 % / 2 -9.9 % / 2 -10.1 % / 2
seq Bull +0.1 % / 33 +0.4 % / 33 +1.0 % / 33 -0.3 % / 33 -0.0 % / 32 +0.1 % / 32
seq_w Bear +0.5 % / 40 +2.6 % / 40 +3.3 % / 40 +5.6 % / 40 +6.6 % / 40 +7.7 % / 40
multi Bear -1.3 % / 1 -0.3 % / 1 -4.4 % / 1 -4.7 % / 1 -8.7 % / 1 -10.0 % / 1
seq Bear -0.0 % / 27 +0.3 % / 27 +0.9 % / 27 +1.4 % / 27 +0.8 % / 27 +0.2 % / 27
Number that Hit +/-10.0% gain/loss in bull/bear by:
Signal Direction 1 Bar 5 Bar 10 Bar 15 Bar 20 Bar 25 Bar
-------- ----------- ------- ------- -------- -------- -------- --------
seq_w Bull 0 / 35 0 / 35 0 / 35 0 / 35 0 / 35 0 / 35
multi Bull 0 / 2 0 / 2 0 / 2 0 / 2 0 / 2 0 / 2
seq Bull 0 / 33 1 / 33 3 / 33 4 / 33 5 / 32 8 / 32
seq_w Bear 0 / 40 0 / 40 0 / 40 0 / 40 0 / 40 2 / 40
multi Bear 0 / 1 0 / 1 0 / 1 0 / 1 0 / 1 1 / 1
seq Bear 0 / 27 0 / 27 0 / 27 1 / 27 2 / 27 3 / 27
Of course, in the last example a bearish signal warning of an 8% drop sounds impressive, until you notice that the bullish signal leads to an even bigger loss and on more data-points.
We can add a control check, which just fires randomly, to give a sort of background-level of price-changes to compare against.
./pyPriceAgent.py -b 2000 -c seq,seq_w,multi,ctrl
Across this tiny data-set, a simultaneous weekly and a daily sequential-9 bull signal led to loss of 3.9% after fifteen days, with four data points.
The equivalent bear signal led to a 0.4% less.
Just randomly picking a time led to a 4.3% gain on a bearish signal and a 2.6% gain on a bullish signal.
What are we to make of this?
Probably that the data-set is too small to mean anything, there's eight data-points and six control points.
From: example@gmail.com
To: example@your.email.moo
Subject: Daily Stock Summary
2020-02-10
Nice looking things this time:
2: TSCO.L 1067 days ago: SEQ Red 9 d, SEQ Red 9 w
2: DBCBTC.HUOBI 46 days ago: SEQ Red 9 d, SEQ Red 9 w
1: BTCUSD.BINANCE 55 days ago: SEQ Red 9 d
1: DGBBTC.BITFINEX 40 days ago: SEQ Red 9 d
1: LBCBTC.BITTREX 1996 days ago: Random Bull
1: AMZN.O 55 days ago: SEQ Red 9 d
1: BTCEUR.COINBASE 1951 days ago: Random Bull
1: IBM.N 126 days ago: SEQ Red 9 d
And some things looking bad:
-1: BTCUSD.BINANCE 186 days ago: SEQ Green 9 d
-1: DGBBTC.BITFINEX 1981 days ago: Random Bear
-1: LBCBTC.BITTREX 1970 days ago: Random Bear
-1: DBCBTC.HUOBI 1972 days ago: Random Bear
-1: BTCEUR.COINBASE 1800 days ago: Random Bear
-1: IBM.N 104 days ago: SEQ Green 9 d
-2: TSCO.L 201 days ago: SEQ Green 9 d, SEQ Green 9 w
-2: AMZN.O 770 days ago: SEQ Green 9 d, SEQ Green 9 w
Average Gain After Signal:
Signal Direction 1 Bar 5 Bar 10 Bar 15 Bar 20 Bar 25 Bar
-------- ----------- ------------ ------------ ------------ ------------ ------------ ------------
seq_w Bull -0.2 % / 99 -1.4 % / 99 -2.7 % / 99 -3.5 % / 99 -5.1 % / 99 -6.3 % / 99
multi Bull +0.2 % / 4 -0.3 % / 4 -1.3 % / 4 -3.9 % / 4 -5.0 % / 4 -5.1 % / 4
seq Bull +0.1 % / 108 +0.2 % / 108 +0.7 % / 108 +0.1 % / 107 +0.1 % / 105 +0.3 % / 105
ctrl Bull +2.0 % / 3 -0.8 % / 3 +3.3 % / 3 +2.6 % / 3 +1.8 % / 3 +0.9 % / 3
seq_w Bear +0.4 % / 154 +1.7 % / 154 +4.2 % / 154 +5.7 % / 154 +6.7 % / 154 +8.7 % / 154
multi Bear +0.1 % / 4 +0.2 % / 4 -0.9 % / 4 -0.4 % / 4 +1.2 % / 4 +2.3 % / 4
seq Bear -0.0 % / 112 +0.4 % / 112 +0.8 % / 112 +1.0 % / 112 +1.4 % / 112 +0.9 % / 111
ctrl Bear +1.0 % / 3 +5.8 % / 3 +4.6 % / 3 +4.3 % / 3 +3.6 % / 3 +4.0 % / 3
Number that Hit +/-10.0% gain/loss in bull/bear by:
Signal Direction 1 Bar 5 Bar 10 Bar 15 Bar 20 Bar 25 Bar
-------- ----------- ------- ------- -------- -------- -------- --------
seq_w Bull 0 / 99 0 / 99 0 / 99 0 / 99 0 / 99 0 / 99
multi Bull 0 / 4 0 / 4 0 / 4 0 / 4 0 / 4 0 / 4
seq Bull 0 / 108 4 / 108 8 / 108 11 / 107 12 / 105 17 / 105
ctrl Bull 0 / 3 0 / 3 0 / 3 0 / 3 0 / 3 1 / 3
seq_w Bear 0 / 154 0 / 154 0 / 154 0 / 154 0 / 154 2 / 154
multi Bear 0 / 4 0 / 4 0 / 4 0 / 4 0 / 4 1 / 4
seq Bear 0 / 112 2 / 112 3 / 112 8 / 112 15 / 112 17 / 111
ctrl Bear 0 / 3 0 / 3 0 / 3 0 / 3 0 / 3 0 / 3
Often you may see a signal in your daily report and wonder how it will turn out. We can tell pyPriceAgent to remind us when there's a result on a position we take, or are just curious about but not confident enough to bet on.
We saw that the RSI signalled bearish on BTCUSD, so we figure it's going to go down. We can place a bet:
./pyPriceAgent.py -t BTCUSD -B X/9500/+5%/20/30
PyPriceAgent will tells us that it understood we wanted to bet that the price of BTCUSD would go from it's last close down to 9500, but that if it went up by five percent (to 10321.50, it calculated) then you'd call it a loss.
You'd also call it a day at 20 days, regardless of how things looked.
Placing Bet: 2020-02-10, BTCUSD.BINANCE
->From 9851.23 To 9500.00 with a stop of 10343.79 or 20 days (Conf: 30.00)
When the script runs each night after close from then on it'll check if your bet has won or lost or timed out yet and add that to your daily report when it happens.
You can use X to get a default on any field, and use percents or direct numbers in the ones where that makes sense.
./pyPriceAgent.py --help
| NAME | ALERTS WHEN |
|---|---|
| bets | alert the end/outcome of bets placed [default] |
| ctrl | fire randomly, as a control check |
| emasort | ema sort-order changes [default] |
| emax1 | short ema crosses [default] |
| emax2 | med ema crosses |
| emax3 | long ema crosses |
| emax4 | vlong ema crosses [default] |
| masort | ma sort-order changes |
| max1 | short ma crosses |
| max2 | med ma crosses |
| max3 | long ma crosses |
| max4 | vlong ma crosses |
| multi | look for multiple other-check hits [default] |
| rsi | daily RSI crosses 30/70 [default] |
| rsi_w | weekly RSI crosses 30/70 [default] |
| seq | daily sequential 9 candles [default] |
| seq_w | weekly sequential 9 candles [default] |