Fractional shares are not primarily a feature for small investors who cannot afford a full share of an expensive stock. They are a precision requirement for any rebalancing system that wants to hit target allocations accurately. Without fractional share support, a 500 EUR portfolio with 20 positions cannot achieve its target weights — the rounding error introduced by whole-share constraints pushes actual allocations 8–15% away from targets before the market has moved at all.
This piece explains why whole-share constraints break rebalancing precision, what the actual error magnitude looks like at different portfolio sizes, and what platforms need from their execution layer to solve the problem.
The arithmetic of whole-share constraints
Consider a simple scenario. A 500 EUR portfolio targets a 5% allocation to a position in a Xetra-listed ETF that trades at 72 EUR per share. The target position value is 25 EUR (5% × 500 EUR). At 72 EUR per share, the exact quantity needed is 0.3472 shares.
With whole-share constraint, the platform can buy 0 shares (under-allocating by 25 EUR) or 1 share (over-allocating by 47 EUR — buying 72 EUR of a position that should represent 25 EUR). Neither is close to the target. The error on this single position is either 100% of the intended allocation (buying zero) or 188% of it (buying one).
At 500 EUR with 20 positions averaging 25 EUR each, every position with a share price above approximately 25 EUR faces this constraint. European equity ETFs on Xetra and Euronext typically trade in the 20–120 EUR range. The majority of positions in a diversified 20-position portfolio will face some degree of whole-share rounding error at small portfolio sizes.
The problem is sharper for individual equities and certain high-price ETFs. A Borsa Italiana-listed Italian bank share at 28 EUR creates moderate rounding pressure at 500 EUR portfolio size. A Nasdaq Stockholm-listed industrial holding at 185 SEK equivalent (approximately 16 EUR at current rates) creates less pressure, but a 500 EUR portfolio still can only hold 31 shares rather than a fractional 31.25. Over a 20-position portfolio, these individually small rounding errors aggregate to a portfolio that is systematically misallocated from construction.
Error magnitude by portfolio size
The whole-share rounding error as a fraction of intended position value diminishes as portfolio size increases. At a 500 EUR portfolio with 20 positions, the average rounding error per position runs 15–30% of the intended position size. At 5,000 EUR (25 EUR target per position scaled to 250 EUR), the average rounding error falls to 1.5–3% per position. At 50,000 EUR, it is 0.15–0.3% per position — comparable to normal bid-ask spread costs.
For European retail neobanks, where many users start with small initial investments (100–1,000 EUR), the portfolio population most affected by whole-share constraints is also the largest population in user count. These are the investors for whom accurate target-weight execution is arguably most important, because they have least capacity to absorb allocation errors relative to their investment goals.
We are not saying that whole-share constraints make rebalancing useless for small portfolios. A 500 EUR portfolio with 8% rounding error per position is still better managed than no rebalancing at all. The argument is that the cost of fixing the problem — implementing fractional share support — is low compared to the systematic misallocation it prevents.
How rounding error compounds over rebalancing events
The problem is not just at initial portfolio construction. Rounding error compounds through rebalancing events because each rebalance starts from an already-imprecise allocation and attempts to correct it with whole-share trades.
Imagine a portfolio that starts 8% underweight its target equity ETF position because whole-share constraints prevented the initial allocation from being accurate. A drift-triggered rebalance fires when the actual deviation exceeds the threshold. The rebalancing calculation determines that 11.4 EUR of additional buying is needed. With whole-share constraint, the closest available trade is either 0 EUR (buy nothing) or 72 EUR (buy one full share). Neither corrects the deviation precisely. The portfolio oscillates around an allocation that is persistently wrong in ways that are hard to measure and harder to explain to investors.
With fractional share support, both the initial construction and each subsequent rebalancing event can achieve the target allocation to four decimal places. The 0.3472 share buy executes. The 0.1583 share sell executes. The portfolio tracks its target allocation continuously, with drift that reflects genuine market price movements rather than arithmetic rounding.
Fractional shares across European exchanges
Fractional share execution in European retail brokerage is not uniformly available across all venues. The infrastructure for fractional execution — which involves the broker splitting a whole share between multiple client accounts and tracking fractional ownership — is implemented at the broker level, not at the exchange level.
In practice, the major retail-facing brokers in Europe (and the custodians that serve neobanks) have implemented fractional share support for the most liquid exchange-traded products on the primary venues: Euronext, Xetra, and LSE. Coverage on less liquid venues (Borsa Italiana, Wiener Börse, some Nasdaq Nordic listings) is more variable and broker-dependent.
For a platform building rebalancing products, this means the fractional share question cannot be answered purely at the API level — it depends on what the custody layer supports. The rebalancing engine must be able to handle both fractional and whole-share execution paths per instrument, and must flag instruments where fractional execution is not available so that the rebalancing plan can account for residual rounding.
The API parameter fractional_precision controls the decimal places used in quantity calculations. Setting it to 4 (the default) means quantities are computed to four decimal places. The execution system then rounds to the nearest supported precision for the relevant instrument and venue. The residual — the difference between the computed fractional quantity and the executed quantity — is tracked and carried forward to the next rebalancing event.
What residual tracking means in practice
When fractional execution is not available for an instrument, the rebalancing engine must track the unexecuted residual and include it in the next plan. If a position requires buying 2.7500 shares and only whole-share execution is available, the engine executes 2 shares and records a residual of 0.7500 shares still needed. At the next rebalancing event, the calculation starts from the current position (2 shares) plus the residual (0.7500 shares pending) to determine the incremental trade needed.
Without residual tracking, each rebalancing event re-derives the needed quantity from scratch. On a position where the target weight is small relative to the share price, successive whole-share rounds repeatedly calculate a fractional quantity that rounds down to zero — the position never gets bought. The portfolio runs persistently underweight the position, which creates real tracking error that compounds over time.
Residual tracking converts the fractional-share problem from a structural deficiency into a managed approximation: the portfolio won't hit its target on every rebalancing event, but the error is bounded and will be corrected as positions accumulate to sizes where the whole-share constraint is no longer binding.
Tax lot implications of fractional shares
Fractional share support interacts with tax lot tracking in a way that platform teams sometimes overlook. When a position is built up through multiple fractional-share purchases — 0.3472 shares in January, 0.2918 shares in March, 0.4110 shares in June — the tax lot record must track these as distinct lots with distinct acquisition prices, just as it would for whole-share purchases. The fractional quantities do not merge into a single averaged lot.
This matters for the tax-optimal lot selection discussed elsewhere, and it matters for the bed-and-breakfasting 30-day rule under TCGA 1992 in UK GIA accounts. Selling 0.35 shares of a fractional position and re-buying within 30 days triggers the same wash-sale rules as a whole-share transaction. The rebalancing engine's TCGA tracking cannot assume that fractional quantities are below a threshold that exempts them from wash-sale calculation — no such threshold exists in the legislation.
For most rebalancing scenarios, this interaction is manageable: fractional sells of small positions do not typically generate large enough gains or losses to create material wash-sale exposure. But platforms should ensure their lot tracking schema supports fractional quantities at full precision (four or more decimal places) rather than rounding lot quantities to whole numbers at the storage layer.
Practical recommendations for platform teams
If your platform serves users with account sizes below 5,000 EUR, fractional share support is not optional — it is the difference between a rebalancing product that works and one that systematically misallocates capital.
Three practical steps:
1. Understand what your custody layer actually supports. Ask your custody or prime brokerage provider specifically which ISINs and which exchanges support fractional execution. Get a definitive list, not a vague "we support fractional on most products." The instruments where fractional is not available are exactly the ones where the rounding problem is most acute.
2. Implement residual tracking from day one. If you are building rebalancing infrastructure, build residual tracking into the data model early. Adding it retroactively requires migrating historical position records to include a residual field, which is more complex than doing it at the start. The residual field should be per-ISIN per portfolio, not a portfolio-level aggregate.
3. Surface the rounding impact in simulation mode. Before running live rebalancing, run simulation mode on your portfolio population to quantify the actual allocation error from whole-share constraints. If the simulation shows that 40% of your portfolios have rounding errors greater than 5% on at least one position, that is a signal that fractional execution support is critical for your portfolio size distribution. If the simulation shows errors below 1% everywhere, whole-share constraints are not a material problem for your current user base.
Fractional shares are infrastructure. They do not appear in marketing materials or compliance documents. But for any European retail platform running rebalancing at scale across small-balance accounts, the difference between fractional precision and whole-share approximation is the difference between a product that delivers what it promises and one that chronically underperforms its stated objectives.
The arithmetic is unambiguous. The solution is available. Platforms that have not yet built it should prioritise it.