Business Analytics
Final Group Paper
Overview
In real life scenario, the case has been inspired by John Orta, a team member who is
currently investing in cryptocurrencies and other team members, Luis Perez Armas and
Shereene Mou Tham, who are curious about mining. Therefore, the main objective of this case
is to analyze whether we should invest in crypto coins or mine them.
History of Bitcoin
Pre-bitcoin years began in 1998. Though the concept was formed at that stage, it was
not fully developed yet. In 2008, the real beginning of Bitcoin happened when a certain Mr.
Nakamoto posted Bitcoin to a mailing list discussion on cryptocurrency. From then on, the
breakthrough of Bitcoin has shown developments each passing year. The Bitcoin software was
made available to the public in 2009. A year later, Bitcoin was valued for the first time and in
another year rival cryptocurrencies started emerging.
Case Analysis
Today everybody is talking about bitcoin price and cryptocurrencies, people speak about
their electronic wallets and the amazing benefits of blockchain, how this new source of money
Trade or mine?
Do we trade cryptocoins or we mine them?
What is the best GPU to mine cryptocoins?
Which cryptocoin to mine?
The initial question is weather to trade cryptocurrencies or to mine them, to answer this
question the best tool available is to set up a decision tree, using as a variable for decision the
profit from mining cryptocurrencies in different scenarios of price, and compare it with the same
revenues coming from trade on the same scenarios.
Since it is almost impossible to test all the 1500 coins available in the market, we
selected five of the currently most popular, with different maturities: Ethereum ETH, Ethereum
Classic ETC, BitcoinGold, Monero and Zcash, all of them have different price ranges, and
different difficulties to mine. We gathered the current prices and set up 5 different scenarios:
High (+100%), Good (+20%), Reg (+0%, or the current price), low (-50%) and crash (complete
loss of value) Table 1. We also compiled the data of daily production of the six most popular
GPUs for mining the 5 selected coins Table 2.
Coins Normal (Current) High (+100%) Good (+20%) low (-50%)
Ethereum $ 665.40 $ 1,330.80 $ 798.48 $ 332.70
Zcash $ 298.91 $ 597.82 $ 358.69 $ 149.46
EthereumClassic $ 22.14 $ 44.28 $ 26.57 $ 11.07
BitcoinGold $ 86.44 $ 172.88 $ 103.73 $ 43.22
Monero $ 275.63 $ 551.26 $ 330.76 $ 137.82
For the case, we are assuming an initial investment on GPUs of 2000 dollars, and
calculating the resulting revenues for the production of cryptocurrencies of each one of the
selected GPUs and coins, at the different pricing scenarios, we are also calculating the costs
of electric consumption of each GPU assuming a unique power consumption of 200W and an
average price of electricity of 0.206 dollars per KWh, we are also assuming that the machines
will produce all the 365 days of the year. Different GPUs have different prices, since some of
them are considered as cheap and other ones are high quality and expensive, therefore it is
expected that with the same initial investment, we can afford a different number of GPUs, prices
and numbers of cards purchased with 2000 dollars can be found on table 3. We will also
assume that at the end of 3 years of usage the cards can be sold for half of the price, therefore
depreciation of equipment’s its calculated for three years up to half of the investment, resulting
in 333,33 dollars per year.
Price one GPU #Cards
1050Ti 205.99 9.709209185
1060 349.95 5.715102157
1070 534.9 3.739016639
1070Ti 579.99 3.448335316
1080 709.22 2.819999436
1080Ti 1019.99 1.960803537
The revenue of each combination of GPU and coin can be obtained using the following
equation:
𝑅𝑒𝑣𝑒𝑛𝑢𝑒 . ($) = (𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 . ) ∗ (𝑃𝑟𝑖𝑐𝑒 ) ∗ (365 𝑑𝑎𝑦𝑠) ∗ (#𝐺𝑃𝑈 )
The cost of production for each one of the GPUs can be obtained using the following equation:
24 ℎ
𝐶𝑜𝑠𝑡 ($) = #𝐺𝑃𝑈 ∗ (200𝑊) ∗ (365 𝑑𝑎𝑦𝑠) ∗ (0.206 𝐷𝑜𝑙𝑙𝑎𝑟𝑠 𝑝𝑒𝑟 𝐾𝑊ℎ)
1000
Profit for each one of the possible combinations can be calculated using the following equation:
𝑃𝑟𝑜𝑓𝑖𝑡 . = 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 . ($) − 𝐶𝑜𝑠𝑡 ($) − 𝐷𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛(𝑦𝑒𝑎𝑟)
ETH
1050Ti 1060 1070 1070Ti 1080 1080Ti
High 1341.779 2595.929 2854.429 2606.604 1249.03 1408.514
Good -733.336 597.1617 1038.231 931.6023 207.9878 303.6784
Reg -1252.11 97.46992 584.1821 512.8519 -52.2727 338.3162
low -2549.06 -1151.76 -550.941 -534.024 -702.924 -663.053
Crash -800 -700 -600 -500 -400 -300
ETC
1050Ti 1060 1070 1070Ti 1080 1080Ti
High 798.8878 2032.705 2175.459 2041.725 911.6212 1118.315
Good -1059.07 259.2277 630.8497 592.675 5.542626 129.5588
Reg -1523.56 -184.142 244.6973 230.4126 -220.977 193.2165
low -2684.78 -1292.57 -720.684 -675.244 -787.276 -735.603
Crash -800 -700 -600 -500 -400 -300
BITGOLD
1050Ti 1060 1070 1070Ti 1080 1080Ti
High 1116.56 1998.695 2891.102 3032.095 3077.264 2123.211
Good -868.468 238.8215 1060.235 1186.897 1304.928 732.4966
Reg -1364.72 -201.147 602.5188 725.5978 861.8443 695.6647
low -2605.37 -1301.07 -541.773 -427.651 -245.865 -484.379
Crash -800 -700 -600 -500 -400 -300
Zcash
1050Ti 1060 1070 1070Ti 1080 1080Ti
High 1026.751 2088.427 2719.631 2858.506 2892.243 2240.413
Good -922.353 292.6606 957.3525 1082.744 1193.916 802.8178
Reg -1409.63 -156.281 516.783 638.803 769.3339 754.2657
low -2627.82 -1278.63 -584.641 -471.048 -292.121 -455.078
Crash -800 -700 -600 -500 -400 -300
Monero
1050Ti 1060 1070 1070Ti 1080 1080Ti
High -134.19 818.8297 1398.477 1263.841 745.8484 737.4501
Good -1618.92 -469.098 164.6601 125.9448 -93.9211 -98.9601
Reg -1990.1 -791.08 -143.794 -158.529 -303.863 2.784148
low -2918.05 -1596.03 -914.929 -869.714 -828.719 -830.819
Crash -800 -700 -600 -500 -400 -300
23%
Good
1304.93
1304.93 1304.93
23%
NVIDIA 1080 BitcoinGold Reg
3 861.844
= 1084.588025 = 1084.588025 861.844 861.844
23%
Bad
-245.865
-245.865 -245.865
10%
Crash
-400
-400 -400
23%
Good
400
400 400
23%
Trade Reg
0
0 115 0 0
23%
Bad
-1000
-1000 -1000
10%
Crash
-2000
-2000 -2000
23%
High
-134.19
-134.19 -134.19
23%
Good
-1618.92
-1618.92 -1618.92
23%
Monero Reg
-1990.1
= -1578.7835 -1990.1 -1990.1
23%
Bad
-2918.05
-2918.05 -2918.05
10%
Crash
-800
-800 -800
D
ataF
ileC
rypto.xlsx Decissiontree.xlsx
This information was used to calculate the revenues, cost and expected profits on three
different scenarios of productivity, a high or good scenario on which the GPU will be working
30 days of the month, an average scenario on which the GPU will be working 22 days of the
month and a bad scenario on which the GPU will be working 15 days of the month, this
productivity refers to the reliability of the system and a probability of scenario was assigned to
each one of the countries, Table 7.
The revenue for each one of the countries was calculating using the raw daily data of
productivity of each one of the GPUs but only for BitcoinGold (Table 2, line 4), for each one of
the scenarios, using the following equation, on a monthly basis, assuming an initial investment
of the same 2000 Dollars:
𝑅𝑒𝑣𝑒𝑛𝑢𝑒 . = (𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑣𝑖𝑡𝑦 ) ∗ (#𝐺𝑃𝑈 ) ∗ (𝑃𝑟𝑖𝑐𝑒 ) ∗ (𝑊𝑜𝑟𝑘𝑖𝑛𝑔𝑑𝑎𝑦𝑠 )
The cost of production was calculated on a monthly basis, for each one of the GPUs on
each one of the countries and for each one of the scenarios, using the following equation:
24
𝐶𝑜𝑠𝑡 ∗∗ = (𝐶𝑜𝑠𝑡 𝑜𝑓 𝐸𝑛𝑒𝑟𝑔𝑦 ) ∗ (𝑃𝑜𝑤𝑒𝑟𝐶𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 ) ∗ (𝑊𝑜𝑟𝑘𝑖𝑛𝑔𝑑𝑎𝑦𝑠 ) ∗ (#𝐺𝑃𝑈 ) ∗ ( )
1000
Profit was calculated for each one of the scenarios on each one of the countries and for
each one of the GPUs using the following equation:
𝑃𝑟𝑜𝑓𝑖𝑡 ∗∗ = 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 . − (𝐶𝑜𝑠𝑡 ∗∗ )
The results of Revenue, Cost and profit on each scenario and for each country the
calculations can be found on the following table 8:
Venezuela PW Good Av Bad BTCgold day Good Av Bad Good profit Avg profit Bad Profit
1050Ti 75 8.388757 6.151755 4.194378 0.0081 203.9411622 149.5569 101.9706 195.55241 143.405097 97.7762
1060 96 6.320446 4.634994 3.160223 0.0122 180.8089156 132.5932 90.40446 174.48847 127.958211 87.24423
1070 150 6.461021 4.738082 3.23051 0.0194 188.1027482 137.942 94.05137 181.64173 133.203933 90.82086
1070Ti 140 5.561475 4.078415 2.780738 0.0212 189.5751306 139.0218 94.78757 184.01366 134.943347 92.00683
1080 200 6.497279 4.764671 3.248639 0.0249 182.0892812 133.5321 91.04464 175.592 128.767468 87.796
1080Ti 265 5.985941 4.38969 2.992971 0.0281 142.8816361 104.7799 71.44082 136.8957 100.390176 68.44785
US PW Good Av Bad BTCgold day Good Av Bad Good profit Avg profit Bad Profit
1050Ti 75 193.99 142.2593 96.995 0.0081 203.9411622 149.5569 101.9706 9.9511627 7.2975193 4.975581
1060 96 146.1603 107.1842 73.08015 0.0122 180.8089156 132.5932 90.40446 34.648607 25.4089784 17.3243
1070 150 149.4111 109.5681 74.70555 0.0194 188.1027482 137.942 94.05137 38.691643 28.3738718 19.34582
1070Ti 140 128.6091 94.31335 64.30456 0.0212 189.5751306 139.0218 94.78757 60.966017 44.7084122 30.48301
1080 200 150.2496 110.183 75.12478 0.0249 182.0892812 133.5321 91.04464 31.839711 23.3491216 15.91986
1080Ti 265 138.4249 101.5116 69.21244 0.0281 142.8816361 104.7799 71.44082 4.4567496 3.26828302 2.228375
France PW Good Av Bad BTCgold day Good Av Bad Good profit Avg profit Bad Profit
1050Ti 75 99.61649 73.05209 49.80824 0.0081 203.9411622 149.5569 101.9706 104.32468 76.5047624 52.16234
1060 96 75.05529 55.04055 37.52765 0.0122 180.8089156 132.5932 90.40446 105.75362 77.5526561 52.87681
1070 150 76.72462 56.26472 38.36231 0.0194 188.1027482 137.942 94.05137 111.37813 81.677293 55.68906
1070Ti 140 66.04252 48.43118 33.02126 0.0212 189.5751306 139.0218 94.78757 123.53261 90.5905826 61.76631
1080 200 77.15518 56.58047 38.57759 0.0249 182.0892812 133.5321 91.04464 104.9341 76.9516708 52.46705
1080Ti 265 71.08305 52.12757 35.54152 0.0281 142.8816361 104.7799 71.44082 71.798586 52.6522966 35.89929
Table 8. Profit, revenue and cost for each one of countries, GPUs and Scenarios
On the previous analysis the result of the decision tree showed that the best option would
be to go for an expensive GPU, however this result was obtained by assuming the same power
consumption 200W, but when we make an analysis of the behavior of price, unitary productivity,
total productivity and power consumption this result might vary, Table 9.
Card Consumption Price # Cards Productivity 1 Productivity T
1050Ti 75 205.99 9.709209185 0.0081 0.078644594
1060 96 349.95 5.715102157 0.0122 0.069724246
1070 150 534.9 3.739016639 0.0194 0.072536923
1070Ti 140 579.99 3.448335316 0.0212 0.073104709
1080 200 709.22 2.819999436 0.0249 0.070217986
1080Ti 265 1019.99 1.960803537 0.0281 0.055098579
When we look at this results normalized and we see the trend of the behavior we would
notice that the cost of production and price, increases at a higher rate for the more expensive
cards than the total productivity, which might suggest that a cheaper GPU, that consumes less
energy but in a higher quantity will product a better profit result than an expensive GPU that
consumes plenty of energy Table 10, Figure 4.
Card Consumption Price Productivity 1 Productivity T
1050Ti 100% 100% 100% 100%
1060 128% 170% 151% 89%
1070 200% 260% 240% 92%
1070Ti 187% 282% 262% 93%
1080 267% 344% 307% 89%
1080Ti 353% 495% 347% 70%
Table 10. Behavior of Normalized variables for each GPU mining BitcoinGold
All the results calculated on profit were used to build up a decision tree, that would help
us to determine which decision results better regarding weather to use cheap equipment’s or
expensive ones and to set up operations on a cheap energy country which is unreliable or on
an expensive reliable. The results obtained on the decision tree suggests that the best option
is to set up operations on Venezuela using NVIDIA 1050Ti Figure 5. The worst possible decision
is the complete opposite which is setup operations in the US using an expensive GPU NVIDIA
1080Ti, figure 6.
5%
Good
195.55
195.55 195.55
25%
Venezuela Average
143.4
0 114.0665 143.4 143.4
70%
Bad
97.77
97.77 97.77
Figure 5. Best decision on country GPU
15%
US Average
3.26
0 4.16 3.26 3.26
5%
Bad
2.22
1080Ti 2.22 2.22
1
Note: Data, calculations and the decision tree can be consulted by clicking on the following excel
icons:
As per the result the most important outcomes will be the NPV, the Static Payback and
the IRR. In order to quantify appropriately the risk involved on this business certain variable will
be set as a distribution, these variables are: Price, Production rate day and the cost of energy.
10 Thousand simulations are run in order to find out the distribution of the results in NPV, IRR,
Static Payback and a possible outcome of profits coming from trading instead of mining on a
scenario that we choose to invest the same 10 Thousand dollars on buying cryptocoins and
trading them each month at a different price.
Results obtained from the simulation on figure 7, suggest a positive NPV in average of
1649 dollars with a standard deviation of 1030 dollar, and we are 96% sure that such an
investment in mining will produce a positive value at a discount rate of 1.25% per month. The
results obtained from the static payback figure 8, suggest a static payback in average of 2.53
years, with an standard deviation of 0.74 years, and we have a 91% confidence level that we
will recover the initial investment in less than 3 years, based on the free cash flows from the
operations.
The results obtained of the simulations figure 9 suggest what we assume, since the price
might behave randomly, the final output in profit selling and buying randomly result in almost a
0 average profit scenario, although when we look at the standard deviation, it has a value of
7167 dollars, this means that an scenario on which we can obtain a 20 thousand dollars profit
with an investment of 10 thousand is possible, with low probability but possible, however we
can still loss all of our investment as well. If we look with more details in the result an scenario
of the same profit from mining 1649 dollars has a probability of only 41% compared with the
96% obtained from mining which has less capacity for high profits but it is more secure.
Price Productivity
10 12 14 16 18 20 22 24 26 28 30
$ 50.00 6.30 5.69 5.19 4.76 4.41 4.10 3.83 3.60 3.39 3.21 3.04
$ 60.00 5.65 5.06 4.58 4.18 3.85 3.56 3.32 3.11 2.92 2.75 2.60
$ 70.00 5.12 4.55 4.10 3.72 3.41 3.15 2.93 2.73 2.56 2.41 2.28
$ 80.00 4.68 4.14 3.71 3.36 3.07 2.83 2.62 2.44 2.28 2.15 2.02
$ 90.00 4.31 3.79 3.38 3.06 2.79 2.56 2.37 2.20 2.06 1.93 1.82
$ 100.00 3.99 3.50 3.11 2.81 2.55 2.34 2.16 2.01 1.88 1.76 1.66
$ 110.00 3.72 3.25 2.88 2.59 2.35 2.16 1.99 1.85 1.72 1.61 1.52
$ 120.00 3.48 3.03 2.69 2.41 2.18 2.00 1.84 1.71 1.59 1.49 1.40
$ 130.00 3.27 2.84 2.51 2.25 2.04 1.86 1.71 1.59 1.48 1.38 1.30
$ 140.00 3.09 2.68 2.36 2.11 1.91 1.74 1.60 1.48 1.38 1.29 1.21
$ 150.00 2.92 2.53 2.23 1.99 1.80 1.64 1.51 1.39 1.30 1.21 1.14
Figure 12. Static payback vs Days of production and Coin Price Area
Figure 13. Static payback vs Days of production and Coin Price Area
Price Productivity
10 12 14 16 18 20 22 24 26 28 30
$ 50.00 -2.06% -1.85% -1.64% -1.43% -1.23% -1.02% -0.82% -0.62% -0.43% -0.23% -0.04%
$ 60.00 -1.83% -1.58% -1.33% -1.08% -0.83% -0.59% -0.36% -0.12% 0.11% 0.34% 0.57%
$ 70.00 -1.61% -1.31% -1.02% -0.73% -0.45% -0.17% 0.10% 0.37% 0.64% 0.90% 1.16%
$ 80.00 -1.38% -1.05% -0.72% -0.39% -0.08% 0.24% 0.55% 0.85% 1.15% 1.44% 1.73%
$ 90.00 -1.16% -0.79% -0.42% -0.06% 0.29% 0.64% 0.98% 1.32% 1.65% 1.97% 2.30%
$ 100.00 -0.94% -0.53% -0.13% 0.27% 0.65% 1.03% 1.41% 1.78% 2.14% 2.49% 2.85%
$ 110.00 -0.73% -0.28% 0.16% 0.59% 1.01% 1.42% 1.83% 2.22% 2.62% 3.00% 3.38%
$ 120.00 -0.52% -0.03% 0.44% 0.91% 1.36% 1.80% 2.24% 2.67% 3.09% 3.50% 3.91%
$ 130.00 -0.31% 0.21% 0.72% 1.22% 1.70% 2.18% 2.64% 3.10% 3.55% 4.00% 4.43%
$ 140.00 -0.10% 0.46% 1.00% 1.53% 2.04% 2.55% 3.04% 3.53% 4.01% 4.48% 4.95%
$ 150.00 0.11% 0.70% 1.27% 1.83% 2.38% 2.91% 3.44% 3.95% 4.46% 4.96% 5.45%
Table 16. NPV, IRR, Payback vs Cost of Energy & markdown price on assets
Note: Data, calculations and the simulations can be consulted by clicking on the following excel icons:
Sources
https://www.forbes.com/sites/bernardmarr/2017/12/06/a-short-history-of-bitcoin-and-crypto-
currency-everyone-should-read/2/#6b1ce3fc533c
https://www.coindesk.com/information/how-bitcoin-mining-works/
http://si-journal.org/index.php/JSI/article/viewFile/335/325