Portfolio experiment

Aktiesparekonto, started 20.07.2026, day 119. Numbers from the account every 15 minutes.
Value
62,457 kr
17.11.2026 17:30 UTC
Since start
-1.14%
ACWI
+0.78%
vs ACWI
-1.92pp
NowPortfolio 62,457 kr -1.14%ACWI, same start 63,673 kr +0.78%
Show as a table
DayPortfolio krSince startACWI same start krSince start
17.11.202662,457-1.14%63,673+0.78%
16.11.202663,899+1.14%65,035+2.94%
13.11.202664,300+1.77%65,315+3.38%
12.11.202664,846+2.64%66,023+4.50%
11.11.202664,703+2.41%65,769+4.10%
10.11.202663,523+0.54%64,516+2.12%
09.11.202663,444+0.42%64,396+1.93%
06.11.202662,661-0.82%63,713+0.85%
05.11.202662,998-0.29%63,813+1.00%
04.11.202663,947+1.22%65,295+3.35%
03.11.202663,440+0.41%64,711+2.43%
02.11.202663,824+1.02%65,115+3.07%
30.10.202663,985+1.28%64,970+2.84%
29.10.202664,650+2.33%65,959+4.40%
28.10.202664,792+2.55%66,138+4.68%
27.10.202664,567+2.20%66,053+4.55%
26.10.202664,108+1.47%65,415+3.54%
23.10.202663,736+0.88%65,100+3.04%
22.10.202663,329+0.24%65,145+3.11%
21.10.202664,189+1.60%65,889+4.29%
20.10.202663,742+0.89%65,554+3.76%
19.10.202663,980+1.27%65,694+3.98%
16.10.202664,141+1.52%65,544+3.74%
15.10.202664,069+1.41%65,794+4.14%
14.10.202664,496+2.08%66,393+5.09%
13.10.202664,506+2.10%66,428+5.14%
12.10.202664,618+2.28%66,742+5.64%
09.10.202664,274+1.73%66,548+5.33%
08.10.202663,835+1.04%66,228+4.83%
07.10.202664,151+1.54%66,483+5.23%
06.10.202664,869+2.68%66,817+5.76%
05.10.202664,461+2.03%66,543+5.32%
02.10.202663,970+1.25%65,834+4.20%
01.10.202663,390+0.33%65,205+3.21%
30.09.202663,759+0.92%65,375+3.48%
29.09.202663,916+1.17%65,175+3.16%
28.09.202664,052+1.38%65,110+3.06%
25.09.202664,235+1.67%65,345+3.43%
24.09.202664,157+1.55%65,025+2.92%
23.09.202664,342+1.84%65,464+3.62%
22.09.202664,311+1.79%65,674+3.95%
21.09.202664,204+1.62%65,255+3.29%
18.09.202663,838+1.04%64,347+1.85%
17.09.202664,150+1.54%64,496+2.09%
16.09.202663,765+0.93%64,042+1.37%
15.09.202663,527+0.55%63,623+0.70%
14.09.202663,894+1.13%63,877+1.11%
11.09.202663,997+1.30%64,227+1.66%
10.09.202663,467+0.46%63,618+0.70%
09.09.202663,764+0.93%63,947+1.22%
08.09.202663,980+1.27%64,496+2.09%
07.09.202663,971+1.25%64,566+2.20%
04.09.202663,902+1.14%64,601+2.25%
03.09.202664,034+1.35%64,656+2.34%
02.09.202663,707+0.84%64,187+1.60%
01.09.202663,587+0.65%64,057+1.39%
31.08.202663,529+0.55%64,152+1.54%
28.08.202663,761+0.92%64,836+2.62%
27.08.202663,257+0.12%64,272+1.73%
26.08.202663,511+0.53%64,077+1.42%
25.08.202663,452+0.43%63,952+1.22%
24.08.202663,478+0.47%63,678+0.79%
21.08.202663,718+0.85%63,897+1.14%
20.08.202663,452+0.43%63,743+0.89%
19.08.202663,639+0.73%64,077+1.42%
18.08.202663,934+1.20%64,207+1.63%
17.08.202664,583+2.22%65,105+3.05%
14.08.202664,533+2.14%64,990+2.87%
13.08.202664,458+2.03%65,255+3.29%
12.08.202664,413+1.95%64,925+2.76%
11.08.202664,354+1.86%64,791+2.55%
10.08.202664,223+1.65%64,781+2.54%
07.08.202663,765+0.93%64,781+2.54%
06.08.202663,795+0.98%64,591+2.24%
05.08.202663,640+0.73%64,621+2.28%
04.08.202663,724+0.86%64,436+1.99%
03.08.202663,369+0.30%63,478+0.47%
31.07.202662,989-0.30%62,610-0.90%
30.07.202662,408-1.22%62,201-1.55%
29.07.202662,513-1.05%62,420-1.20%
28.07.202662,497-1.08%62,934-0.39%
27.07.202662,577-0.95%62,924-0.40%
24.07.202663,002-0.28%63,179+0.00%
23.07.202662,736-0.70%62,754-0.67%
22.07.202663,077-0.16%63,658+0.76%
21.07.202662,553-0.99%63,578+0.63%
20.07.202663,167-0.02%63,179+0.00%
20.07.202663,179+0.00%63,179+0.00%

Strategy and status

Strategy and status

In short (11.11.2026 12:03 UTC): the account is in the index and stays there. Eight tests today, every rule written before its run. Failed: buying after a directed share issue, the theme question on new jumps, orders and raised figures without a jump, buying a theme fund once it runs, and a sleeve that always holds the strongest theme funds (4 points a year behind the world index). Held as a filter: the compound rule's questions sort jumps on 3.177 new cases, but since 2021 the good ones only match the index. One lead with support: the questions work in the least traded shares and not in the big ones, in Sweden and, on the cases there are, in Denmark and Finland. That is now a second paper rule, small shares only, from tonight. Nothing found reaches your bar of 1 point a month; no theme or sector fund did either, even picked with hindsight.

16.11.2026 08:05 UTC: the whole account is in the index. R3 done: 140 IUSN sold at 9,25 EUR, 99 WEBN bought at 13,09 EUR (Saxo fills), 220 DKK cash left.

14.11.2026 10:12 UTC: the index only. The research now looks for one niche where the swings are big for us and too small for big players, and goes deep there.

What is in force

RuleWhatUntil
R1Move into the indexDone 09.11.2026
R3Sell all IUSN, buy WEBN (your word 14.11)Done 16.11.2026
R2Theme slice: up to 5% on an HN theme call, your yes per buy31.12.2026
S1Strategy research, no fixed proposal date (your word 10.11)31.12.2026
S2Ping you when a call is worth money31.12.2026
S3Daily status ping, asks only on real money31.12.2026

Default (your word, 14.11.2026): the index only. Everything not earned by a tested edge sits in WEBN (655 units since 16.11.2026).

The goal

Beat ACWI after all costs, with a model improved in versions (v1, v2 ...). A version gets real money only by the ladder 5% → 10% → 25% → more, each step on a forward record.

Target (10.11.2026): a separate account in your company, about 500.000 DKK, once a model seems to work. This account is the test. The model is built and paper-traded at that size and with that account's costs; the direction is small Nordic companies with slow signals (insiders, steady buyers) and warnings as exits.

What the research has shown

When to buy (07-09.11): no edge. Tested against random days for the same shares, after costs:

Which shares to hold (10.11, first results):

The niche, first tests (10.11.2026, Sweden, 574 shares, 2005-2026):

Big moves: what we know so far (10.11.2026 13:50 UTC)

Sweden, 07.2016-08.2026, Saxo prices, exploratory.

GroupCasesMeanMedianAhead of the indexDoubled / halved
Checklist says buy22-3,2%-5,3%60 / 0
Checklist says pass18-13,6%-17,0%42 / 3
All 459, no reading459+0,3%-6,1%39%20 / 25
  1. As a buy rule it fails. Most jumps are a good report in an ordinary company. Those neither double nor halve; they drift a little behind the index.
  2. As a warning it works. The pass group lost 14 points and held every halver. On all 457 jumps I could check, a jump with no company news behind it lost 8,8 points to the index (median -12,5%), in each of the three years.
  3. Both doublers were in the pass group. The big winners come out of the same broken or unfunded shares as the halvers.
  4. The one jump that had everything, a theme included, was the best buy: MilDef, February 2025, +59% against the index. One case, so a lead, not a result.

What I am watching, and why

Forward data recorded from 09.11 by timers (backup nightly):

The crowd gauge (10.11.2026 19:33 UTC)

How much Avanza's customers trade (trades a day per 1.000 customers, from Avanza's monthly statistics), against its own three-year median.

The compound rule, on paper from 11.11.2026 (10.11.2026 20:09 UTC)

Your idea: stack several things. Each new jump gets four yes-or-no questions:

  1. Is the crowd in? (the gauge above)
  2. Is it a sound jump? Real company news behind it, no share issue, the share at its high.
  3. Is there a booster? An order, raised figures, or an insider who bought.
  4. Is its theme running? A big Hacker News story on the theme in the last 45 days.

What jumps did on history by the number of yes answers to the first three (60 trading days against the world index, 2016-2026):

Yes answersJumpsAgainst the indexHalved
0534-6,8%8,1%
1669-0,3%5,7%
2201+4,2%0,5%
385+9,8%1,2%
Yes answers (first three)Jumps2016-20202021-2026
01.326-1,6%-5,6%
2 or 3499+3,5%-0,1%
3115+9,1%+3,7%
  1. The questions sort. About 5 points between the bottom and the top, in both halves. That part is real, not fitted.
  2. The level is not there. Since 2021 the top group only matches the index. Small Swedish shares as a group have trailed the world index since 2021, and the rule does not lift them over it.
  3. Orders and raised figures are the steady question (5 points in both halves). The crowd question is weaker here.
Jumps with the yes answers2016-20202021-2026
Sweden, shares under 1 million SEK a day, big jumps (the rule's own history)+20% (34)+11% (60)
Sweden, same shares, smaller jumps (new cases)+10% (52)+3% (52)
Denmark and Finland, same size of share (new markets)too few (15)+10% (33)
Sweden, shares trading 5 million SEK a day or more (big jumps, smaller jumps)+5%, 0%-5%, -2%
  1. Jumps without the yes answers in the same small shares lost 6 to 11% since 2021. None of the 198 Swedish cases with the yes answers halved.
  2. I found the pattern after the fact in Sweden, then tested it on Denmark and Finland with the rule written first. There the half that can be judged agrees; the other half has too few cases. So: support, not proof.
  3. The typical case is near zero in Sweden (a few winners carry the average) and a trade in such a share costs about 4%.
  4. It is now paper rule version 2 (small shares only), fixed before its first trade; about 20 buys a year on history, fewer while the crowd gauge is off. It fits this account's size; a 500.000 DKK account could only trade the upper end of these shares.

Theme funds as the trade (11.11.2026 11:27 UTC)

Your plan is the index plus big opportunities. A big theme can be bought in this account as a fund. I took every theme and sector fund on the list of Xetra, the German exchange, that Saxo offers: 282 funds (chips, defence, banks, uranium, robotics and so on), prices from 2012, against the world index.

Removing the losers: screened index funds (14.11.2026 09:58 UTC)

You asked whether research that removes the many losing companies could beat the index a little. I tested the version anyone can buy: every world fund at Saxo that screens out losers (quality, value, momentum, multifactor, low volatility), against the plain MSCI World fund, as held, after fees, in EUR. The rules were fixed before any price was fetched.

Research log, last runs (UTC)

Where this might land (10.11.2026 18:55 UTC)

A simulation on 2016-2026: the account in the world index, with 5% positions in the jumps that pass today's leads (order news in a share at its high, a trading update, insider buying), 60 days each, after costs.

Next

No fixed date (your word, 10.11: research first). The attack list (11.11.2026 12:03 UTC):

  1. Tonight: the first scored scan of both paper rules, after the Stockholm close.
  2. Denmark and Finland every day. Next I would set up the daily scan for Danish and Finnish small shares, so the small-share rule covers them, and read the 48 Danish and Finnish cases one by one.
  3. Avanza owner inflow per share, once a few weeks of our own data exist.

Closed today: the directed issue as a buy, orders and raised figures without a jump, theme funds bought on strength. On hold: paid data (ask A12), the crowd gauge abroad, shorting. The 15% in small caps stays, and moves into the index only if the research ends without a model (your word, 10.11).

Last decision

14.11: IUSN into the index and the index only from now on; the research hunts for one niche where big swings are too small for big players (your idea). IUSN trailed the world index by 3,0 a year since 2018 (test T30).

09.11: account moved into the index (sold Equinor, Danske Bank, BNP Paribas; bought 198 WEBN). Next decision: none dated; a proposal comes when the research supports one.

Written by the trading agent, last changed 16.11.2026 18:20 UTC.

Positions

InstrumentQtyOpenLastCurP/L DKK%Value DKK
Amundi Prime All Country World Acc UCITS ETF655.0012.7912.73EUR-294-0.5%62,189
Total-29462,189

Machine

Disk free
99.1 GB
477.0 GB / 600.6 GB used (79%)
Database (bolig)
198.6 GB
RAM available
26.4 GB
4.2 GB / 30.6 GB used
Swap used
0 B
of 8.0 GB
CPU now
0%
load 0.3 / 0.4 / 0.5 · 16 cores
Uptime
3d 14h
Load is the 1 / 5 / 15-minute average; on this box anything under the core count is comfortable. Measured 2026-11-17 17:30 UTC.
Rendered 17.11.2026 17:30 UTC from 5110 snapshots. Prices are 15 minutes delayed. Not investment advice.